Phaedo

It seems preferable to take the present lines either (i) as restating the idea of 87d8—e5 that the death of the man might be due to the perishing of his soul, or (ii) as recalling the idea of 88a9—b3 that one of the soul's deaths might 'bring perishing' to it. With (i), 'death is the perishing of soul' will be an explanation rather than a redefinition of death. Death is due to the perishing of soul, somewhat as a black­out is due to a power-failure. The cessation of vital functions is due to a breakdown of their source. On this view, the words simply resume Cebes' original idea of soul as an agent of bodily change. With (ii), 'death is the perishing of soul' will mean that the trauma of separation from the body will sooner or later prove fatal to the soul. This would link the words closely with 88al0—b2, and 'death' would be used in precisely the sense there specified. It is possible that both these ideas are in Socrates' mind, but the emphasis on constant bodily flux (cf.91d7 with 87d9) strongly suggests that he is thinking at least of (i). If he is thinking only of (i), there is no reference here to Cebes' concession at 88a4—7 regarding reincarn­ation. That will be glanced at only in the second restatement of his position (95d4-6).

On any interpretation of these lines 'death' must, here as at 88al0—b4 and 95d4, mean the 'event' of dying rather than the 'state' of being dead. See note 4, and on 71d5—e3. For the concept of 'death', see also on 57al-b3, 64c2-9, 88al-b8, 88cl-89cl0, 105c9—dl2, 105dl3-e9, 105el0-107al (p.221).

92a6-c3. Socrates here points out that the attunement theory of the soul conflicts with the belief in its prenatal existence, which Simmias had accepted. This point is ad hominem, and would, of course, be ineffective against someone who rejected the Recollection Argument. But how effective is it against someone who accepts it? As noted above (see on 86b5—d4), the attunement theory may be variously interpreted. An attunement may be (a) a ratio, (b) a tuned state, or (c) music. On any of these interpretations, Simmias might reply that an attunement could, in fact, have existed before the particular body in which it inhered. Thus, (a) the ratio governing the lengths of lyre strings could be said to exist before any given lyre; (b) the tuned state of lyres in general could be said to exist before some particular lyre came into existence; and (c) music, e.g. a scale or melody, could exist before any given instrument on which it might be played. In short, neither tunings nor tunes depend for their existence upon that of particular instruments.

Simmias makes no such response, however. To do so would require 'attunement' to be understood in such a way that different lyres could share the same attunement. On such a view, Simmias' theory would entail that different bodies could share the same soul. To avoid this consequence, it must be supposed that the attunement of each instrument is unique to it, and numerically distinct from the attunements in all others. It could not then exist before its own lyre, or survive that instrument's destruction. Understood thus, the attunement theory would make Simmias' original point, and would also be defeated by Socrates' present objection.

At 92b7—8 Socrates says: 'attunement isn't, in fact, the same kind of thing as that to which you liken it' (sc. the soul). This interpretation follows Burnet's text, and assumes a slight looseness in the use of 'likening', since it is, strictly, the soul that has been likened to an attunement, and not vice versa. For this reason Verdenius prefers a variant reading, and would replace 'as that to which you liken it' with 'as you represent it' or 'as you guess it to be'. But if Simmias can be said to have 'represented' an attunement as being of any particular nature, it is only as being liable to perish before its component elements, and Socrates is not disputing that.

This note, and subsequent notes on Socrates' refutation of the attunement theory, owe much to an unpublished paper by Mr. C. C. W. Taylor.

92cll—e3. Simmias now renounces the attunement theory as incompatible with the theory of Recollection, which was in turn derived 'from a hypothesis worthy of acceptance' (d6—7, cf.el—2), i.e. the Theory of Forms. It is plausible to see his withdrawal as an application of the 'hypothetical method' that Socrates will describe at 100a3—7, the positing of the theory he judges to be strongest, and the taking of things not in accord with it to be untrue. Admittedly, Simmias has not yet been told of the hypothetical method. But it is because he believes in adopting 'the best and least refutable of human doctrines' (cf.85c8—dl with 100a3—4) that he now has to retract the attunement theory. He is committed, unawares, to Socrates' own method. See on 84d4-85dl0.

The sense of the words translated 'just as surely as its object exists—the Being, bearing the name of "what it is"' (d8—9) is uncertain. Loriaux (155, cf. E.F.P. 31) argues that the phrase 'bearing the name' must have a causal nuance, and embodies a proof of the Forms' existence. This, he thinks, dictates the interpretation 'in virtue of its bearing the name of that which is"—i.e. 'that which exists'. The present version follows Loriaux in taking the main verb of the 'just as' clause (the first 'is' at d9) existentially. But the sense of 'is' in the 'what' clause cannot be fixed as he argues. To derive the Forms' 'being' from the nomenclature used to refer to them would be a singularly weak 'proof of their existence. See also note 50.

92e4—93al0. Socrates here begins a new assault on the attune­ment theory. The section that follows, 93all-94b3, is extremely difficult, and its analysis remains highly problematic. Reference should be made to Archer-Hind, Burnet, Hackforth, Verdenius, Bluck, and W. F. Hicken, C.Q. 1954, 16-22. Olympiodorus' commentary (ed. W. Norvin, 169) and Philoponus on Aristotle De Anima 1,4. (ed. M. Hayduck, 141-5) are also relevant.

The argument of 93all— 94b3 will be referred to here as 'Argument B'. It is followed at 94b4-95a3 by a further argument against the attunement theory, which will be called 'Argument A'. Steps in each argument will be numbered accordingly (A1 etc., B1 etc.). The attunement hypothesis itself will be referred to as H.

The admissions of 92e4—93al0 serve as premisses for Argument A. At 93al 1 Socrates appears to start afresh, eliciting a new set of admissions (93all—clO), to be used as premisses for Argument B. The structure of the whole passage 92e4—95a3 is therefore as follows:

Premisses for Argument A (92e4-93al0);

Premisses for Argument B (93al 1-clO);

Argument B (93dl-94b3);

Argument A (94b4-95a3).

Here the pattern A-B-B-A is well marked (cf. Hicken, op.cit. 17). Yet it is hard to see why Plato has cast the arguments in this form. If the A premisses are left unused until Argument A, the train of thought begun by them is interrupted for more than a page. Archer-Hind (78) provides continuity by linking the A with the B premisses: the admission at 93a6—10 that an attunement is governed by its constituent elements is supposed to support the claim at 93all—b3 that the extent of every attunement depends upon the extent to which it is tuned. But this seems unlikely. The words 'Again now', which preface the B premisses at 93all (cf.92e4,94b4), serve to divide them from the A premisses rather than to link them. Nor, in fact, are these sets of premisses logically connected. The dependence of an attunement upon its constituent elements does not imply its dependence upon the extent to which it is tuned. The former dependence concerns the state of an attunement (al), how 'it acts and is acted upon' (a4), whereas the latter concerns the extent to which it is an attunement (93b 1—2). Moreover, the constituent elements are not mentioned in the B premisses at all. The extent to which an attunement is an attunement is said to depend not upon how its elements are tuned, but upon how it is tuned itself (93al2, al4).

The A and B arguments, and their respective premisses, are therefore better taken as logically unrelated. Further discussion of the A premisses will be deferred until Argument A is considered. See on 94b4-95a3.

93all—b3. The B premisses begin here. Socrates asks (i) 'Isn't it natural for every attunement to be an attunement just as it's been tuned?' (all—12). Simmias fails to understand, and is next asked (ii) 'Isn't it the case that if it's been tuned more and to a greater extent, assuming that to be possible, it will be more an attunement and a greater one, whereas if less and to a smaller extent, it will be a lesser and smaller one?' (al4—b2).

What difference, if any, is intended between 'more' and 'to a greater extent', and between 'less' and 'to a smaller extent'? (2) Does this passage imply that degrees of attunement are possible, or that they are not, or does it imply neither of these things?

For (1), which is of minor importance, see note 51. Whatever difference may be intended between the two pairs of terms, no use is made of it in the argument. For simplicity, therefore, only the antithesis between 'more' and 'less' will be used in these notes.

, however, is critical for the argument. Burnet takes the passage as an admission that no attunement admits of degree. Socrates first asks (i) whether the nature of every attunement depends upon how it has been tuned, i.e. whether it varies according as it has been tuned to the interval of a fourth, a fifth, or an octave. He then suggests (ii) that if, per impossibile, an attunement were more or less tuned, it would not be more or less an attunement. On this view, (ii) would have to be understood as asking: 'And it isn't the case, is it, that (even) if it has been tuned more, it will be more of an attune­ment, whereas if it has been tuned less, it will be less of one?' However, if framed in this way, Socrates' question would expect the answer 'No', and Simmias' 'Certainly' at 93b3 would be inappropriate. Moreover, on Burnet's interpretation, (ii) would be an inept reply to Simmias' 'I don't understand.' For it would not explain what Socrates has asked in (i), but would make a completely fresh point. Yet (ii) must be meant to clarify for Simmias what he had been asked in (i) (cf. W. F. Hicken, C.Q. 1954, 19). It must, therefore, be suggesting that if an attunement has been more or less tuned, assuming that to be possible, it must be more or less an attunement. The character­ization of the attunement will depend upon that of the tuning process. The first premiss of the argument may therefore be interpreted thus:

Bl. If an attunement has been more or less tuned, assuming that

to be possible, it will be more or less an attunement.

What is meant, however, by the proviso 'assuming that to be possible'? Burnet says that this plainly indicates that it is not possible for an attunement to be more or less tuned, hence that it could not be more or less an attunement than another. However, the 'assuming' clause need not, in itself, imply 'per impossibile''. Socrates does not say 'if, in fact, that were possible', but 'if, in fact, that is possible'. Cf.93e8 and 94a9, where 'assuming' introduces supposedly true premisses of the argument.

If Socrates were talking about the attunement of an instrument, Burnet's view would be clearly mistaken, for different degrees of tuning in a lyre .obviously are possible. However, the subject of the sentence is 'attunement', and it is not easy to attach sense to an attunement's being tuned in different degrees, or even at all. Moreover, there would seem no point in adding the proviso 'assuming that to be possible', if the possibility were obviously beyond dispute. It therefore seems preferable to suppose, with Hicken, op.cit. 20, and Bluck, 100, n.l, 198, that the question whether an attunement can be tuned more or less is here simply left open. The argument will then turn neither upon its possibility (Philoponus, Archer-Hind, Hackforth), nor upon its impossibility (Burnet). A quite different account of the proviso given by Verdenius (see note 52) leaves the argument equally unaffected.

93b4—7. The next premiss is:

B2. One soul is not, even in the least degree, more or less a

soul than another. Burnet takes this as derived from the previous admission together with the hypothesis that soul is attunement. The present sentence certainly could be translated as an inference, but has not been so taken here. That one soul is not more or less a soul than another seems hard to deny, and would be ill supported by the far more questionable proposition that attunement admits of no degrees (cf. W. F. Hicken, C.Q. 1954, 19). But in any case, there are serious objections to taking the latter proposition as a premiss (see previous note).

Philoponus, partially followed by Hackforth (118), takes 93all— b7 as a self-contained argument. Thus:

An attunement can be more or less an attunement;

A soul cannot be more or less a soul;

Hence,

A soul cannot be an attunement.

However, (i) (1) has not been asserted (see the account of B1 in the previous note), (ii) Such an argument would be defective. Attunement as such might admit of degrees, while specific kinds of attunement, such as souls, did not. As Philoponus says, the argument would have to run 'every attunement admits of degrees' (cf. Hicken, op.cit. 18, n.l). Moreover, (iii) if Simmias' theory were already refuted at 93b7, Socrates might be expected to say so, before passing on to a fresh point. 'Well, but' at 93b8 is therefore better taken as continuing with a further phase of the same argument (cf. the same phrase at 79b 1), than as introducing a new one.

93b8—clO. Two further premisses are now added:

B3. Some souls are good, and others bad (b8—c2).

B4. Good souls contain attunement, and bad souls contain non-

attunement (c3—8). Both premisses are accepted as self-evident, and must be assumed to be stronger than the attunement theory itself, if the argument is to succeed as a reductio ad absurdum of it. The denial of B3 will constitute the absurdity to which the theory will be reduced (94a8— 10).

B4 suggests an account of goodness and badness like that of the Republic (430e, 443d—e), where temperance and justice are defined in terms of attunement within soul and state. Socrates will argue here that such an account would be incompatible with Simmias' attunement theory. But would it be any more compatible with his own earlier suggestion that the soul is 'incomposite'? The account of goodness in the Republic is based upon a doctrine of 'parts' of the soul. If attunement or non-attunement can be contained only by that which is composite, they would require a different account of the soul from that which the Phaedo suggests. Plato recognizes and tries to resolve the conflict between the doctrine of a composite soul and that of immortality at Republic 61 lalO—612a7. See also on 78cl—9.

Philoponus finds in 93c3—8 a further self-contained argument against the attunement theory: taken together with B3 and B4, it implies that an attunement (a) may have a further attunement, or (b) may have a non-attunement, within itself, either of which would be absurd. It is true that to a modern reader (a) sounds ill-formed, while (b) appears to breach the principle maintained later (102d5— 103c8) that opposites cannot admit each other. Some such objection to (b) might be read into the argument at 94a2—4, but there is no indication that Socrates is appealing to it here. As for (a), the idea of an attunement's having a further attunement within itself seems no more objectionable than that of its being tuned, which figures twice in the argument (93al 1—bl, 93d6—9), yet is never dismissed as improper. Nor is the conclusion required by Philoponus' inter­pretation, that soul is not attunement, actually drawn at 93c9—10. The argument therefore remains, as yet, incomplete (cf. W. F. Hicken, C.Q. 1954, 17,n.3).

93d 1—5. Socrates now reaffirms B2, and then says: 'and this is the admission that no one attunement is either more or to a greater extent, or less or to a smaller extent, an attunement than another'.

Thus:

B5. One attunement is not more or less an attunement than

another (d3-4).

Burnet takes B5 as reiterating 93all—b3, which he interprets as claiming that attunement admits of no degrees. Thus he translates: 'this is just our admission, (namely B5)'. But his interpretation seems untenable for the reasons given above. See on 93al 1—b3. The natural sense of the present lines is 'this (sc. B2) amounts to the admission that B5'. See note 53. On this view, B2 yields B5 by substitution of 'attunement' for 'soul', based on Simmias' hypothesis (H). This raises two questions: (1) Is B5 consistent with Bl? (2) Is it validly derived from B2?

B5 will be inconsistent with Bl, if, but only if, the latter is construed as asserting categorically that an attunement can be more or less an attunement. There need be no inconsistency, if Bl is taken as non-committal regarding this possibility, as urged above (see on 93all—b3). Some, however, find the present lines irreconcilable with 93all—b3, and would therefore emend the text at 93d4. For the effect of the emendation upon the argument, and some objections to it, see note 53.

Those who would retain the MS. reading (e.g. Archer-Hind, Hicken) sometimes construe B5 as restricted to the attunements specifically under discussion, namely soul-attunements. Nowhere in the text is any such restriction stated. But if B5 extends to attune­ments in general, or refers to 'attunement' as such, its derivation from B2 is fallacious. From the premiss that no soul is more or less a soul than any other, conjoined with the hypothesis that soul is an attune­ment, it would not follow that no attunement whatever is more or less an attunement than any other, but only that no soul-attunement is more, or less a soul-attunement than any other.

It should be noted that Greek lacks the indefinite article, and the sense of 'is' in 'soul is attunement' (H) is therefore unclear. In English we may distinguish more easily between (HI) 'soul is (identical with) attunement' and (H2) 'soul is an (or a kind of) attunement'. Socrates' refutation of H depends upon interpreting it as HI, and treating 'soul' and 'attunement' as interchangeable salva veritate. Substitutions of'soul' and 'attunement' are made, according­ly, at 93d 1—4, 93dl2-e2, and 94a2-6. The attunement theory is, indeed, often expressed simply as 'soul is attunement' (92c9—10, 92e3, 93c3, 94a2, 94bl-2, 94c4, 94e2-3). But Simmias clearly need not be committed to the view that 'soul' and 'attunement' are equivalent terms. It is clear from several passages (86c2—3, 88d4—5, 91dl—2, 92a8, 94e8-95al) that his meaning would be better expressed by H2. If so, Socrates' refutation of him is open to the

above criticism.

93d6—e6. The argument continues:

B6. That which is not more or less an attunement has not been more or less tuned (d6—8).

B7. That which has not been more or less tuned participates in

attunement to an equal degree (d9—11).

Hence,

B8. Soul has not been more or less tuned (dl2—e3). Hence,

B9. Soul does not participate more in non-attunement or in attunement (e4—6).

With Burnet's text at 93d4, the subject 'that which' (d6, d9) in B6 and B7 must be 'the attunement' rather than 'the soul', as would be required if the text were emended—see note 53. B6 then follows from B1 by contraposition. B7 introduces the new termin­ology 'participating in attunement', which seems to mean 'being in a state of attunement' as distinct from 'being an attunement'. If so, it is best taken as an independent premiss, rather than (with W. F. Hicken, C.Q. 1954, 21) as derived by conversion from B6: an equal degree of tuning entails an equal state of attunement. For the distinction between 'being in a state of attunement' and 'being an attunement' see next note.

B8 is prefaced by a repetition of B2, but is presumably derived by substituting 'soul' for 'attunement' in B6. B9 is derived from B8, together with a similar substitution in B7. The point of repeating B2 in obtaining B8 is not clear, but it may be meant to recall its role at B5, and consequently to suggest resubstitution of 'soul' for 'attune­ment' in B6 and B7. Strictly, however, what is needed to obtain B8 and B9 is the reintroduction of H. For the validity of substitutions based upon H, see previous note.

93e7—94b3. The attunement theory is now reduced to absurdity as follows:

BIO. One soul does not participate more in goodness or badness than another (e7-10).

Bll. A soul could never participate in badness (94al—7). B12. All souls of all living things are equally good (a8—11). B13. B12 conflicts with B3 (supplied). Hence,

B14. H is not correct (al2-b3).

Here BIO is derived from B9 by substituting 'goodness' for 'attunement' and 'badness' for 'non-attunement', on the strength of B4. B11 is given as a still more paradoxical conclusion than BIO. B12 is based upon BIO and B11, with further reference to B2.

BIO and B12 are straightforward steps. The derivation of Bll, however, is much more problematical. It is obtained from H (94a2) together with:

B5*. An attunement is completely itself, namely an attunement (a2—3). This yields:

B5**. An attunement could never participate in non-attunement (a3—4).

Bll then follows by substitution of 'soul' for 'attunement' and 'badness' for 'non-attunement', based on H and B4 respectively, in B5**.

How is B5** supposed to follow from B5*? No doubt B5* should be taken as an abbreviated form of B5, 'One attunement is not more.,or less of an attunement than another'. The reasoning will then be from this assertion to B5**. But this transition is highly dubious. For it is arguable that, although every attunement is, indeed, as much an attunement as any other, nevertheless some attunements lack attunement, and some may, to use the terminology introduced at B7, 'participate' in attunement more or less than others. The shift to this terminology was represented in the previous note as a change from 'being an attunement' to 'being in a state of attune­ment'. This distinction must now be clarified.

'Attunement' may be taken to mean either a tuning (attunement¹) or a correctly tuned state (attunement²). It might be agreed that every attunement¹ is an attunement¹ equally, no one attunement¹ more or less so than any other. But it may also be held that some attunements¹ participate in attunement² more or less than others, and that there is no contradiction in holding that an attunement participates in non-attunement², i.e. lacks attunement². Thus it could be admitted that every attunement¹ is equally an attunement¹, yet denied that every attunement¹ is equally in a state of attunement².

For example, the normal tuning of a guitar is E-A-D-G-B-E. This might be held to be no more an attunement¹ than the variant tuning D-A-D-G-B-E, which involves lowering a single string by one tone. It might also be held to be no more an attunement¹ than, say, F-A- DfrG-A#C, a random combination of notes to which the strings could be, but are not in practice, tuned. Either of these latter tunings might be said to be no less an attunement¹ than the usual one. But if the usual tuning is assumed to be the only 'correct' one, then the second and third tunings mentioned could be regarded as participating less in attunement² than the usual one. There would be no contradiction in holding that they participated in non- attunement², or lacked attunement². The third tuning might intelligibly be said to participate in attunement² less than the second, because it is, in an obvious way, further from the normal tuning than is the second. Note that these points are quite inde­pendent of a distinction that might be drawn between any of these tunings and particular instances of them. They could be expressed equally well in terms of the tunings themselves or in terms of their particular instances.

It seems, then, that B5 is defensible if it is interpreted solely in terms of attunement¹. But 'being an attunement¹' does not entail 'being in a state of attunement²'. B5** therefore does not follow from B5, and seems, moreover, actually false. Applying this result to the soul, it may be argued that its being an attunement in the sense required by Simmias' theory need not preclude it from lacking attunement in some sense that would enable good and bad souls to be distinguished in the way proposed at B4. And if the derivation of Bll is fallacious, the further conclusion at B12, that all souls are equally good, will be open to similar objections.

It is hard to be sure that the argument equivocates upon 'attunement' in the way just suggested. But if it does, the root ambiguity is one that pervades the use of many abstract nouns, in both Greek and English, such as 'height', 'length', 'depth', 'size', 'weight', 'thickness', or 'speed'. In sense 1 we may speak of David's height as well as Goliath's, of the tortoise's speed as well as the hare's. But in sense 2 it is only Goliath who has height, and only the hare who has speed. Thus, David's height¹ can lack height²; the tortoise's speed¹ can lack speed²; and a lyre's attunement¹ can lack attunement .

Possibly, in denying that attunement can participate in non- attunement, Socrates would rely on the principle of 102d5—103c8, that 'opposites will not admit each other'. But it will be evident that that principle would not really be infringed by the supposition that an attunement¹ can admit non-attunement². For attunement¹ and non-attunement² are not opposites of each other.

A stronger reply for Socrates to make would be to deny that the attunement theory is correctly represented in terms of attunement¹. Soul, according to the theory, was not just any tuning of the bodily elements, but their correctly tuned state, i.e. attunement². Cf.86cl — 2, 'when they're blended with each other in due proportion'. It therefore remains debatable whether good and bad souls (attune- ments²) could, in fact, be distinguished in terms of a further attune­ment² in the way proposed at B4.

Hackforth (119—20) takes 94a2—5 as withdrawing the assump­tion, provisionally made at 93al4-bl, that one can tune a lyre more or less exactly. But that assumption was neither made nor denied in the earlier passage (see on 93al 1—b3). Nor is it here denied that varying degrees of tuning in a lyre are possible. Bluck (100, n.l) cites Republic 349el0—16 as denying this, although, in fact, it suggests just the opposite. But controversy over whether a lyre can be tuned more or less exactly is largely irrelevant. For the soul is not being compared with a lyre, but with an attunement; and it is with the implications of an attunement's (not a lyre's) being tuned that the whole argument is concerned.

The reductio ad absurdum is finally sprung at 94al2—b3. B13 has been supplied in the summary above, to enable the attunement theory (H) to be represented as generating a contradiction: since B12 conflicts with the common-sense intuition B3 accepted earlier, H has to be withdrawn.

Note that at 94b 1 the attunement theory is called a 'hypothesis' (cf.93cl0). The argument is evidently an application of the 'hypothe­tical method' described at 100a and lOlc-d. The hypothesis that soul is attunement has been shown to lead to 'contradiction', i.e. to a consequence (B12) which conflicts with an earlier admission (B3). It is rejected because its 'consequences' are in discord with each other (cf.l01d4—5). See R. Robinson, P.E.D. 142. Of course, B3 is not itself a logical consequence of H but an independent assump­tion. But cf. Robinson (op.cit. 133): '[A hypothesis] may have conflicting consequences on our standing assumptions, that is, when combined with some of our permanent beliefs.' This is the kind of 'contradiction' involved both here and in the refutation of the attunement theory (Argument A) that follows. See also on92cll— e3.

94b4—95a3. Argument A runs thus:

Al. The soul can control and oppose the bodily feelings (94b4— c2,94c9—el).

Furthermore, as was agreed earlier (92e4—93al0):

A2. An attunement can never be in a state other than that of its

components (92e4—93a3).

A3. An attunement can neither act nor be acted upon in any way different from its components (93a4—5,94c3—8). Hence,

A4. An attunement cannot control or oppose its own compon­ents (93a6—9). Hence,

A5. The soul cannot be an attunement (94e8—95al). Al develops and applies the principle used earlier (80a), that soul 'rules' body. It often opposes bodily inclinations and prevents their gratification. These include anger and fear, hunger and thirst (94b8—

10, d5). All such states are here lumped together as 'bodily', and are viewed as sources of conflict between soul and body. In the Republic Plato will treat similar conflicts as evidence of different 'parts' within the soul itself. The first of the two lines quoted from the Odyssey (xx.17) at 94d8—el is used at Republic 441b6 to support a distinction between 'rational' and 'spirited' elements in the soul. See on 64e4—65a3 (p.89).

A2 and A3, if taken strictly, are overstatements. The attunement of a lyre is clearly capable of 'being acted upon' in a way in which its components are not. For it may be destroyed, while the strings and wood remain intact. Moreover, the strings may be 'acted upon' in such ways as being stretched or severed. These operations, although they affect the attunement, could hardly be regarded as affections of the attunement itself: it is not stretched or severed. Its affections need not, therefore, coincide with those of its com­ponents. But this does not alter the essential point. A lyre's attunement depends wholly upon the state and relationship of its material components, whereas they in no way depend upon it. The causal relation is in one direction only. By contrast, the soul is not only acted upon by the bodily elements, but acts upon them. This is the point at which the attunement theory is being held to break down.

All that would follow from this argument, however, is that the soul is not an attunement of bodily feelings. It would not follow that it is not an attunement at all. To this it might be replied that bodily feelings are, according to the attunement theory, the com­ponents of the soul. But this is not quite in line with Simmias' original account. The components there specified were 'hot and cold, dry and wet, and the like' (86b8—9). Bodily feelings, such as hunger and thirst, or anger and fear, were not mentioned. Yet they are now referred to as 'alleged sources' of the soul's existence (94cl0—dl). No doubt they are to be thought of as due to the presence of the basic physical elements in various proportions, thirst, for example, being associated with heat (94b8—9). But they are clearly not on a par with those elements, and it therefore remains uncertain whether, according to the theory, the 'components' of a soul-attunement are the basic physical elements (as at 86b), or their psychic products (as here), or both.

More generally, the strength of the argument will depend upon whether the phenomena of conflict and self-control demand explanation in Socrates' terms. A defender of the attunement theory might deny that they evidence the activity of an auto­nomous soul. The soul's so-called 'opposition' to the body, he might object, is itself simply the effect of a bodily state.

3.6 Socrates' Story (95a4-l02a9)

Cebes' objection leads Socrates into an account of his own intellect­ual history. He tells of his early interest in natural science, and his abandonment of it in favour of a quest of his own. His method is described and illustrated with reference to the Theory of Forms.

95b5—e6. Cebes' position is here restated once again. As before (88b5—6), it is required that soul be proved 'imperishable and immortal' (cl). Hackforth's view (122, n.l) that these terms are still being used synonymously is rightly rejected by D. O'Brien (C.Q. 1968, 97). But the addition of 'imperishable' does not seem to be dictated, as O'Brien suggests, by Cebes' concession regarding reincar­nation. This concession would, he thinks, allow the soul a limited or 'partial' immortality, i.e. survival of one or more deaths, but would not guarantee it 'full' immortality, i.e. survival of any and every death. 'Imperishability' is therefore added, O'Brien suggests, to fill this potential deficiency in the notion of 'immortality'. See on 91d2—9.

If this were so, we should expect Cebes to allow that, were reincarnation granted, the soul would have been proved 'immortal' in the limited sense, and only its 'imperishability' would remain to be shown. But this is not how Socrates presents Cebes' position here. He argues that the soul's strength, divinity, and prior existence need not prove its immortality (cl, dl); and that it would make no difference whether the soul was incarnate once or many times (d4—5): one would still be foolish not to fear death, unless one could prove the soul immortal (d6—el). A proof of 'immortality' would still be needed, then, even if the soul were reincarnate many times. It follows that 'immortality' cannot be used here in any sense less stringent than the ability to survive every death. And since Cebes agrees that the present resumd expresses his position exactly (e4—6), he can hardly have had anything else in mind at 88a—b.

It is doubtful whether the concept of 'partial' immortality finds any place in the dialogue. At 73a2—3 (see on 72e7—73a3) Cebes had referred to the doctrine of Recollection as showing that the soul is immortal, in virtue of its prenatal existence. But he later complains (77cl-5) that the Recollection Argument has shown only 'half what is needed. It must also be shown that our souls will exist when we have died, 'if the proof is going to be complete'. That objection, essentially, is repeated here (c6—d2), and it is added that it would not be met, even if many incarnations were admitted. All this strongly suggests that for Cebes 'immortality' never connotes any thing.less than 'full' immortality in O'Brien's sense, the capacity to survive any number of deaths without perishing. Nevertheless, there is an important point in the addition of 'imperishable', which will become clear later. See on 105el0— 107al (p.217).

95e7—96a5. Socrates' account of his intellectual history is, as G. Vlastos has said (P.R. 1969, 297), 'one of the great turning points in European natural philosophy'. It is, in fact, a striking counterpart in ancient philosophy of Descartes's Discourse on Method, despite its rejection of 'mechanistic' explanations that Descartes,was to revive. Cf. A. E. Taylor, P.M. W. 200, n.l. Like Descartes, Socrates professes to be confused by the senses and to abandon their use. Both are pioneers of a new philosophical method. Both seek metaphysical foundations for mathematics and natural science. And both formulate basic certainties that fortify their religious convictions. Moreover, the autobiographical form of Socrates' story, as of Descartes's, disguises the true rigour of its author's thought.

For philosophical purposes it hardly matters whether Socrates' story is authentic, and if so, whether it is true of the historical Socrates or of Plato himself, or whether part is true of each. For a balanced review of these alternatives see Hackforth, 127—31. It is more important to ask how if bears upon Cebes' position. Why is his objection here said to require an inquiry into the reason concerning coming-to-be and destruction? This demands an elucidation of two key concepts: (1) 'reason', and (2) 'coming-to-be and destruction'.

(1) The primary sense of the noun translated 'reason' (aitia) is 'charge' or 'accusation'. The related adjective, applied to human agents, means 'responsible' or 'at fault' (cf,116c8), and the cognate verb means 'accuse' or 'blame'. The concept is thus rooted in the notion of human responsibility (cf.98e2—99a4). In a secondary and frequent use, the noun, and the adjective in its neuter form, are applied to a wide range of non-human things, to which events or states of affairs may be attributed. Similarly, the verb can mean 'impute' or 'ascribe' (98b9, c2, el).

In its secondary use the noun has traditionally been rendered 'cause'. This hallowed mistranslation is particularly unfortunate here, since it covers, at most, only part of the field with which Socrates is concerned. Many of the things he will mention are not amenable to what we should call 'causal' explanation. They are not, and could not be, subsumed under causal laws, or related to sets of antecedent conditions sufficient for them to come about. See on 96cl—e5.

Socrates' interest in natural science prompts a number of questions of the form 'why is x FT, which are generally answered by a noun in the dative case ('by y^r) or by a prepositional phrase ('on account of', or 'owing to', or 'because of, y). It becomes ciear from the discussion that these questions mean, initially at least, 'in what does x's F-ness consist?' or 'what constitutes x's F-ness?'. To ask 'why' x is F need not be to ask what 'causes' it to be F, but may rather be to ask for the feature in virtue of which it is F. The range of possible interests covered by Socrates' questions is well reflected in the English question 'what makes x FT, which may embody a request either for causal explanation or for conceptual clarification. Socrates' disenchantment with the natural sciences stems from their failure to pursue the latter. It is not, however, clear that the con­ceptual sense of the question is the only one relevant to the discussion. It is arguable that Socrates will later formulate an aitia that is not constitutive of a thing's being F, but causally imparts F-ness to it, and thus explains how it comes to be F. If such an aitia is provided, then a concern with something like 'causes' in the modern sense is not, after all, to be ruled out. See on 100c9—e4, 105b5-c8 (p.211).

The Protean nature of the concept, and the restrictions that will be placed upon it, make it impossible to find a translation of aitia that fits all of Socrates' multifarious examples, but 'reason' is perhaps the least unsuitable. It has been used for the noun, and for relevant occurrences of the adjective, throughout. See also G. Vlastos, op.cit. 292-6, and E. L. Burge, Phronesis 1971, 1-13.

(2) As noted earlier (see on 70c4—8), the verb translated 'come to be' has both complete and incomplete uses. Similarly, the noun translated 'coming-to-be' (genesis) can mean either a thing's coming- into-being (its 'genesis' in the English sense) or its acquisition of an attribute (see on 71al2~b5). Accordingly, an inquiry into the reason concerning 'coming-to-be' might be either (i) an inquiry into the reason for the coming-into-being of things, or (ii) an inquiry into the reason for their acquisition of attributes. In the ensuing discussion Socrates will confine himself to (ii) (cf. Hackforth, 144— 6). Yet some contribution to (i) is surely demanded from the persp­ective of Cebes' objection. His conception of the soul as a source of vital energy calls for an improved account of things' 'coming-into- being', and not merely of their coming to be, e.g., large or beautiful. Moreover, it is just such an account that Socrates' language leads us to expect. When he asks repeatedly (96a9-10, cf.97b5-6,97c6-7) 'why each thing comes to be, why it perishes, and why it exists', it is natural to take 'comes to be' as the counterpart of 'perishes', i.e. as meaning 'comes into being'. Hackforth (145, n.l) rightly remarks, of the verb cognate with the noun translated 'destruction' at 95e9, that it cannot mean 'lose an attribute'. We should expect, then, that Socrates will explain not merely the acquisition and loss of attributes by already existing things, but the coming-into-being of things that do not exist, and the perishing of things that do.

We find, however, no explicit discussion of these issues. Why are the major questions raised by Cebes' objection apparently dis­regarded? Perhaps because, as Socrates' very first example will suggest (96b2—3), he is concerned primarily with the genesis of living things. And in their case 'to be' is 'to be alive'. 'Coming-into- being' can thus be equated with 'coming to be alive'. Where the value of F is 'alive', 'coming to be F and 'coming to be simpliciter' will coincide. So too will 'ceasing to be F and 'ceasing to be simplicitef, i.e. 'perishing'. A living thing's birth (or conception) is its'coming-into-being', and its death is its 'perishing' or 'destruction'. If these equivalences are assumed, it follows that Cebes' objection demands an inquiry into 'coming to be (alive)', i.e. being born (or conceived), and into 'ceasing to be (alive)', i.e. dying. A conceptual examination is needed of being and coming-to-be, birth and life, death and destruction. This huge task is, as Socrates aptly remarks, 'no trivial matter' (95e8), and his own narrative and discussion are mere prolegomena for it. Apart from a few hints (see on 105el0— 107al, p.220), he does not pursue it himself. He merely offers his 'own experiences' (96a2—3) to Cebes (and the reader) for further reflection.

96bl-8. The question at 96b2—3 seems concerned with the origin and nourishment of the earliest animals, not with the develop­ment of individuals. The theory mentioned, perhaps that of Archelaus (see Burnet's note), conspicuously makes no reference to 'soul'. Similarly, in the next example (b4—5), 'what do we think with?', the answer 'soul' or 'intelligence' is avoided. The verb translated 'think' occurs at 66c5 with different overtones (see on 66b 1—c5). Here it seems to be used in a general way to cover various conscious states from sensation upwards (b5—8). It is perhaps implied that the stock scientific account of perception and thought is inadequate, as providing for no subject of consciousness (cf. Theaetetus 184b—186a and see on 64e4—65a3, p.89).

96cl-e5. G. Vlastos (P.R. 1969, 309, n.50) treats Socrates' 'unlearning' of his former beliefs about nutrition as quite distinct from the 'puzzles' introduced at 96d8—97b3. It marks, he thinks, a preference for 'windy theorising' over homespun explanation, born of Socrates' passion for natural science. But this interpretation destroys what looks like a continuous sequence of thought running from 96c2 to 96e7. At 96c3 Socrates begins to specify what he formerly 'supposed' he knew (c6, cf.d6). He then gives four more examples of what he 'supposed was an adequate view' (d8—e4). When Cebes asks (e5) what he thinks about these things now, he answers (96e6—7) that he is 'far from supposing' that he knows 'the reason' about them. All this suggests a simple contrast between a state of confidence preceding his 'scientific' phase and one of confusion following it, which still, in a manner, persists.

For reasons that will emerge shortly (see next note) Socrates is confused by standard explanations of growth in terms of aggregation. All such explanations, whether couched in simple or sophisticated terms, lead to problems. His claim to have been 'blinded' by science (c5) is ironical (see also on 99d4— 100a3). He pretends that his inability to accept even the simplest common-sense explanation, which rendered him 'unfit' for scientific study, was actually produced by it. In reality, his problems are conceptual. They arise from philosophical reflection, not from empirical study. His pro­fession of 'puzzlement' may seem disingenuous, considering that he has the Theory of Forms up his sleeve (cf. Hackforth, 124, n.2). But the Theory offers only a provisional solution, which is itself in need of further exploration (107b4—10). And Socrates' puzzlement is no doubt meant to be infectious. A reader who first has to puzzle over the nature of his puzzles will come to feel them as his own. See also on 100e5-101b8.

The general nature of these puzzles may be brought out by the following expansion of the text at 96d8—e4, linking the difficulties posed in that passage with the questions of growth at 96c2—7: 'A thing grows, anyone would think, in virtue of its "coming to be large" (d4—5) (i.e. larger than it was. But what does "larger" mean?). Is one man or one horse "larger" than another in virtue of (the measure of the difference between them, namely) "a head"? (If "larger" means "containing a greater number of units of measure­ment", then what does "a greater number" mean?) What makes (a given number of units, such as) ten "greater" than (another number, such as) eight? Is it "greater" in virtue of the accruing to it of two? And is (a given length, such as) two cubits "larger" than (another, such as) one cubit in virtue of exceeding it by half (of itself)?'

Note that Socrates is not seeking a 'causal' explanation of one thing's being larger or more numerous than another, but an account of the concepts 'larger' and 'more numerous'. If this concern is not understood, some of his problems will appear unreal or meaningless. Thus, Hackforth (131) complains that 'there is no more a cause of 10 being greater than 8 than there is of Thursday coming after Wednesday'. Certainly, ten's being greater than eight neither needs nor admits of causal explanation. But this is to miss Socrates' point. In considering 'why' ten is greater than eight, he is interested in what constitutes it as 'greater', what its being 'greater' consists in.

'By a head' (el) is a use of the dative case to express the measure 'by' which one thing exceeds another. It need not be supposed that Plato confuses this dative with a 'causal' one. He is using a trivial example to make a serious point: the difference in size between x and y is not what constitutes x's being larger than y. For a further implication of this key example, see on 100e5—101b8.

96e6—97b7. Some words added to the MS. reading by Wyttenbach and Burnet have not been translated. If they are supplied, an extra 'or the one that's been added' must be inserted after 'that's come, to be two' at 96e9. The words are not essential, but they would enable all possible answers to the question 'what has come to be two?' to be taken into account.

What exactly are Socrates' puzzles? He says that he does not accept the view that when one is added to one, either the latter unit (or, on Burnet's reading, the former), or both units have 'come to be two because of the addition of one to the other*. For the meaning of 'I wonder if at 97a2 see on 62a2-7 (p.79), and cf. H. Reynen, Hermes 1968, 44, n.3. Socrates is asking himself in astonishment (i) whether it is the process of addition that makes them two, and thus (ii) whether any of the things mentioned can properly be said to 'come to be two', i.e. what the subject of the predicate 'come to be two' could be.

These two difficulties are connected. First, Socrates is evidently recognizing that the physical propinquity of two items is not what constitutes their being two. For they were two already, before they were juxtaposed. Cf. Frege: 'Must we literally hold a rally of all the blind in Germany before we can attach any sense to the expression "the number of blind in Germany"?' (The Foundations of Arith­metic, §23). This gives rise to the second problem. 'Two' can be predicated of a pair of items taken together, but of neither taken singly. 'One' can be predicated of either taken singly, but not of both taken together. Cf. Hippias I, 300—303, where number predicates are contrasted with others in this very respect. In the light of this, the predicate 'come to be two' is, indeed, puzzling. For two items taken singly can never be two, and taken together they must always be two. A set cannot, it seems, change its cardinal number, whereas the verb 'come to be...' suggests that it can. More generally, how can things 'come to be F', where F is a character that they either always or never possess? Number predicates appear to be such that things cannot 'come to' acquire them. Comparable difficulties can be seen to arise over the predicates 'come into being' and 'cease to be', if negative existential statements are disallowed. This problem too, inherited by Plato from the Eleatics, lies close to the surface of Socrates' inquiry into 'coming- to-be and destruction'.

A further source of puzzlement is the idea that two opposite processes, addition and division, should be 'reasons' for the same thing, i.e. for one's 'coming to be two' (97a5—b3). The clear implication of this is that two opposites cannot function as 'reasons' for a thing's coming to have one and the same property, i.e. cannot be constitutive of that property. Vlastos (P.R. 1969, 312, n.57) questions both the assumption that opposite processes cannot be 'reasons' for the same state of affairs, and its application to the present case. For, he says, the items that have 'come to be two', by addition and division respectively, are different in either case. But Socrates might reasonably wonder how two different items could acquire a common character in virtue of two opposite processes being performed upon them. If what is sought is a 'reason' that is constitutive of the character in question, then two quite different and opposite answers would be unsatisfactory. The 'safe' answer that Socrates will give (101a—c) will be immune from this criticism. See on 101b9-c9.

97b8—98b6. The translation 'Intelligence' has been used here as best suited to the idea that things are arranged for the best, which Socrates thought implicit in Anaxagoras' theory (97c4—dl, 98a6— b 1), but which 'mind' and 'intellect' fail, in different ways, to convey. 'Intelligence' (nous) should be understood here as a substance term. It is the faculty of thought, or that which thinks, rather than a mental quality, such as 'sagacity' or 'good sense'. It is cognate with the verb translated 'think' (83b 1), used for thinking of Forms, and also with the adjective 'intelligible', used to describe their status as objects of thought (80bl, 81b7,83b4). For the original Anaxagorean theory, see DK 59 B 11, 12,13, 14.

Note that Socrates' first reaction to the theory (97c3) was that it seemed to him 'to be a good thing' that it should be so. What 'seemed a good thing', i.e. what appealed to his own intelligence, was the hope of understanding all things as the work of another intelligence. Accordingly, at 97d7 he says that he was pleased to think that he had found in Anaxagoras an instructor 'to suit my own intelligence'. The translation tries to capture a pun, which is obscured by 'after my own heart' (Hackforth). Burnet says that such a joke would be 'very frigid'. Not only is it entirely in keeping with Socrates' ironical treatment of Anaxagoras, but a repetition of it may be suspected at 98b8—9: 'I beheld a man making no use of his Intelligence at all'. In explaining things, Anaxagoras failed not only to invoke his theory of a cosmic intelligence, but also to use his own.

The programme of explaining natural phenomena in terms of 'what is best' is carried out in detail in the Timaeus, both for the universe as a whole, and for its contents. Cf., e.g., Timaeus 29—34, 44d—46a, 68e-71a, and see G. R. Morrow,P.R. 1950, 147-63. The present passage marks the transition from a mechanistic to a tele- ological conception of the natural order that was to dominate European science for the next two thousand years.

At 97d4 (cf.98b6) the better and the worse are said to be objects of 'the same knowledge'. Socrates may mean that an understanding of one member of a pair of opposites (F) requires an understanding of the other (G). This is clearly the case, where, as with 'better' and 'worse', F and G are overtly relational, so that (x, y) (xFy =yGx). He may also be hinting that evil is a necessary complement to good, and that a complete explanation of the universe would encompass both.

98b7—99b6. In view of the connection between an aitia and the notion of responsibility (see on 95e7—96a5, p. 169), it is natural that Socrates should give, as a paradigm instance of a 'reason', his own judgement that it was better to abide the decision of the Athenian court. In his parody of Anaxagoras he charges him with confusing a reason of that sort with the physical conditions necessary for that reason to take effect. The distinction is related to, though not the same as, the modern distinction between necessary and sufficient conditions. The physical conditions specified are necessary not merely for Socrates' sitting in gaol, but for enabling the true reason to function as a reason (99b2—4). This does not mean that Socrates will regard the judgements of rational agents as the only reasons ever admissible, but only that they are indicated by any appeal to 'intelligence' in explaining something. Nor does Socrates mean that he will countenance only the purposive use of the word 'reason'. He himself will use it later (112b 1) in the sense disparaged here.

Note also: (i) Socrates' account of his reason for staying in gaol (98e2—99a4) contains a wry contrast between morality and exped­iency. By implication, the Athenian court's decision was based purely upon the latter.

Socrates' judgement is in turn attributed to the judgement of the court (e3). Thus, one human decision is explainable by reference to another.

Explanations of action such as Socrates gives may be loosely labelled 'teleological', and linked with Aristotle's 'final causes'. But the reason here mentioned cannot, strictly, be identified with Socrates' 'end'. It is not an object for the sake of which he acts, but consists simply in its seeming to him that the relevant action would be for the best. This judgement needs to be distinguished from the

moral or prudential grounds on which it was based.

(iv) The physical conditions in Socrates' example have their counterparts in the material elements and forces used by the divine craftsman who fashions the universe in the Timaeus. They are 'accessories' or 'co-reasons', necessary for realizing his designs. Cf. Timaeus 46c7-e6, and see G. R. Morrow, P.R. 1950, 151.

99b6-d3. Socrates scolds his predecessors for not acknowledging any supernatural power sustaining the universe, but for thinking to find'an Atlas stronger and more immortal than this', i.e. a permanent material support to hold up the physical world. Their disregard of 'the good or binding' as the real sustaining force points to the theistic account of the cosmos that Socrates himself envisages. Similarly Leibniz, citing the present passage: 'the general principles of physics and mechanics themselves depend upon the action of a sovereign intelligence and cannot be explained without taking it into consideration' (Letter on 'Explaining the Laws of Nature', P.P.L. i.542. Cf. Discourse on Metaphysics, §§ 19—20). The word translated 'binding' (c5) means 'obligatory', and is here connected with 'binding' in the physical sense (cf. Cratylus 418e).

The nature of Socrates' 'second voyage' (c9—dl) has been much debated. The phrase may mean 'taking to the oars when the wind has failed' or 'making a second, safer journey'. The former sense is well attested, and suggests a second-best method of reaching one's destination. Cf. Philebus 19c2—3. But if so, what is Socrates' destination? And in what sense is the approach that he will now describe 'second-best'? For both questions see W. J. Goodrich, C.R. 1903, 381-4, 1904, 5-11, N. R. Murphy, C.Q. 1936, 40-7, L. E. Rose,Monist 1966,464-73.

Socrates' destination is the discovery of 'the reason', i.e. the reason for coming-to-be and destruction. Note that he does not say that he is making a second attempt to find a reason of the kind that he had vainly expected from Anaxagoras. Hence there is no need to interpret the Form hypothesis introduced at 100b in teleological terms, as Bluck and others have supposed. It may simply be taken as an inferior pattern of explanation, containing no reference to the teleological ideal. Relative to that ideal it is, indeed, 'second-best', and the phrase may be read without irony. It would not, of course, be second-best in comparison with the cosmologists whom Socrates has just criticized, and the phrase 'second voyage' would be ironical if related to them. But it need not be so understood. The passage is, rather, in the vein of Simmias' remarks at 85c—d. See on 84d4— 85dl0.

Some commentators would insist that all reference to the teleological ideal must be excluded. See, e.g., G. Vlastos, P.R. 1969, 297, n.15, 302-3, and E. L. Burge, Phronesis 1971, 1, n.2. But this seems to be going too far. Whatever state of mind on this point may be imputed to Socrates, his words do not prove that no allusion is intended by Plato in what follows to teleological explanation. As Vlastos says (303, n.37), both here and in the Timaeus teleological explanations are exemplified solely in the purposeful agency of a mind. But it seems conceivable that the Form hypothesis, and the 'subtler' reason that succeeds or supplements it (105c—d), should point to the action of a divine mind as the 'reason' for coming-to-be. This idea is, in fact, discernible later. See on 105el0— 107al (p.221).

N. R. Murphy (I.P.R. 146-7) holds that Socrates' destination on his 'second voyage' is different from his original one, and thus that he avoids altogether the problems of efficient causation and temporal change. But this would disappoint the expectations aroused by 95e9—96al. Those expectations are, indeed, very imperfectly ful­filled (see on95e7—96a5,p.l71). But to deny any reference whatever to a source of coming-to-be and destruction would be to sever Socrates' story from the objection of Cebes which gave rise to it.

99d4—100a3. This passage has suffered from over-interpretation, especially in the light of the sun simile in Republic vi—vii. See W. J. Goodrich, C.R. 1903, 383—4 for a critique of several mis­understandings. When Socrates says that he had tired of studying 'the things that are' (d4—5), he cannot be referring to Forms, which have not yet entered his narrative, but simply means, non- committally, 'things', i.e. the scientific and mathematical matters raised at 96a—97b, and the conceptual questions arising therefrom. No subtle symbolism need be read into the reference to the sun (d6): its eclipse is mentioned merely as the occasion when people are most inclined to look at it. The 'objects' (e3) that Socrates thought he should not look at with his eyes (el—4) cannot be Forms, for Forms cannot be observed by the senses at all (65d—e, 79a). They must, presumably, be the scientific and mathematical matters mentioned above.

Why did Socrates fear that by using his senses to examine them he might altogether 'blind his soul'? Is he harking back to his earlier, ironical suggestion that scientific studies had 'blinded' him (96c5)? Or is he hinting that such studies, if continued, would have unfitted him for conceptual inquiry? The soul's 'vision', its capacity for 'seeing' Forms, might be thought of as damaged by sensory observ­ation (cf.79c2—9). But there is no allusion to the Forms, at least until Socrates speaks of his resort to 'theories' and of his studying in them 'the truth of the things that are' (e6). For this phrase, which does seem to contain a presentiment of the Form hypothesis, cf.90d6—7. For 'theories' (logoi) at 99e5 and lOOal, see on 90b4—91c5, and next note.

The sentence in which Socrates qualifies his comparison of 'theories' with images (al—3) is confusing in translation. The point is not to deny that 'theories' or concrete things are images, but to question whether the former are more so than the latter. To look at things 'in concrete' is to study them in images at least as much as is looking at them 'in theories'. It is being suggested that 'theories' are images of a higher grade than objects in the sensible world, and thus closer to the Forms. This idea, in a more developed form, governs the structure of Republic ii—ix. See D. Gallop, A.G.P. 1965, 113-31.

'Perhaps my comparison is, in a certain way, inept' (99e6—lOOal): more literally, 'Perhaps that to which I liken things is not like.' The words for 'liken' and 'be like' are cognate with the word translated 'images' (99el, 100a2).

100a3— 9. These difficult lines, together with the further precepts about method at 10Id—e, have been much discussed. See Bluck, 13— 14, 111-12, 160-73, Hackforth, 138-42, R. Robinson, P.E.D., Ch.9, K. Sayre, P.A.M., Ch.l, N. R. Murphy, C.Q. 1936, 40-7, P. Plass, Phronesis 1960, 103—15. The main problems are: (1) What is meant at 100a4 by 'hypothesizing on each occasion the theory (logos) I judge strongest'? (2) How can the metaphor of 'accord' (a5) be interpreted in such a way that 'putting down as true whatever things seem to me to accord with it, and as not true whatever do not' will seem a logically defensible procedure? (3) How. is this procedure related to its context, especially to the illustrations at 100b-101c?

(1) logos has been translated 'theory', so as to leave it open whether it should be taken to mean 'definition' or, more broadly, 'proposition', 'statement'. 'Definition' might seem to fit well with Socrates' account (99el—100a3) of his resort to theories after giving up using his senses. The quest for definitions could be expected to figure in an account of his development. Adoption of what he judged to be the strongest definition in each case would form a natural part of this quest, and would enable him to set down parti­cular things as/% or not F, according as they did, or did not, conform to the definition. This reading would link logos at 100a4 closely with its occurrences at 99e5 and lOOal, where it could well mean either 'definitions' or 'conceptual discussions' aimed at producing them. Bluck (13—14, 164—6) understands the lines in this way.

But this view is untenable. For the meaning of 'the theory I judge strongest' must be gathered, partly at least, from Socrates' illustrations at lOObl—101c9. Bluck would, indeed, avoid this, by taking the latter passage as marking a deliberate shift from Socratic to Platonic doctrine. Socratic definitions are, he thinks, there being supplanted as 'causes' by Platonic Forms. But this interpretation is itself bizarre. Definitions are nowhere said to be 'causes', and the words 'it's nothing new' (lOObl) expressly disclaim the notion that Plato is somehow improving upon what has already been said. Clearly, he is making Socrates elucidate the cryptic utterances of 100a, See R. C. Cross,P.R. 1956,405.

The elucidation consists in (i) 'hypothesizing' that beautiful, good, large, and other Forms exist (b5—7), and (ii) agreeing that particular things are beautiful, large, etc. because they participate in the corresponding Forms. Note that Socrates uses the verb 'hypothe­size' at 100b5 as he did for 'the strongest logos' at 100a3. The latter therefore almost certainly exemplifies what he will later call a 'hypothesis' (101d2, d7, cf.l07b5). For the notions of'hypothesiz­ing' and 'hypothesis', see R. Robinson, op.cit., Ch.7. 'Hypotheses' need not be hypotheses in the modern sense, i.e. explanatory theories as yet unconfirmed. Nor need they be 'hypothetical' in the sense of being conditional in form, though they may need to be supported by further argument (101d3—el, 107b5—8). Only (i), and not (ii), is here explicitly said to be 'hypothesized'. But (ii) is evidently inseparable from (i), being integral to the Theory of Forms itself. And if it is taken to form part of 'the strongest logos', it will explain why the logos is 'strongest'. It is so, in this context, for the same reason that it is 'safe' (100d8—e3): it is proof against certain kinds of counter-argument. It will be convenient to call (i) and (ii) together 'the Form-Reason hypothesis'.

Clearly, neither (i) nor (ii), nor the Form-Reason hypothesis as a whole, amounts to a definition. This precludes logos at 100a4 from meaning 'definition'. Whatever it may mean at 99e5 and lOOal, it must here mean 'proposition' or 'statement'. But the notion of definition is not, in fact, far off. For there is an obvious link between the definitional sense of logos and its use to characterize the Form-Reason hypothesis. In seeking a definition of F, one looks for a feature common to F things, in virtue of which they are F. The question 'What is FT naturally gives rise to the question 'What makes F things FT. The answers to the latter question in terms of Forms at 100b—101c are thus directly relevant to Socrates'interest in the former. If this is borne in mind, no sharp change in the meaning of logos between 99e5 and 100a4 need be supposed. The strongest logos is an appeal to Forms, when other answers to 'what makes F things FT, and thus to 'what is FT, have proved inadequate.

(2) The main difficulty in the notion of 'accord' has been succinctly stated by Robinson (op.cit. 126—9). If it means 'be consistent with', Socrates will be putting down as true whatever propositions are consistent with his theory. But it would seem a quite inadequate ground for putting down something as true that it should merely be consistent with a given theory. On the other hand, if 'accord' means 'is deducible from', Socrates will be putting down as not true whatever propositions are not deducible from his theory. And it would seem an equally inadequate ground for holding something to be untrue that it should not be deducible from a given theory. Hence Robinson concludes that Plato 'does not say quite all that he means'. Although 'accord' cannot, strictly, mean deducibility, what Plato here means is that one should take propositions deducible from one's theory to be true, and those whose contradictories follow from the theory to be untrue.

Hackforth initially (139) takes the notion of 'accord' similarly: 'Any proposition arrived at by what the inquirer deems a valid process of deduction is accepted, and the contradictory of any such proposition is rejected.' But his account of lOld-e, which he believes to give the detail of the process described in the present lines, seems inconsistent with this. He introduces a chain of propositions- G -F-E -D- C -B -A— successively deduced, in that order, from an initial hypothesis H. If these propositions survive the test described at 101d3-5, they will be 'in accord' with each other, and may all be put down as true. But if one of them-F, say-is successfully challenged, then it and G will be at variance: 'the one is not a valid inference from the other' (140). However, Fs not being a valid inference from G is clearly quite different from its being the contradictory of something that is a valid inference from H. And whereas the latter defect would give ground, to one who had adopted H, for putting F down as 'not true', the former surely would not.

As noted in (1) above, a quite different interpretation is favoured by Bluck. He takes 'accord' to mean 'conformity' of things to a definition: 'Having found an "account" or "definition", [Socrates] would accept as genuine instances of the thing concerned whatever seemed to conform to it, and reject what did not' (111, cf.l 14, n.l). But it seems hardly credible that the Greek at 100a5 should mean 'accept as genuine instances of the thing concerned'. The word that Bluck translates 'genuine' is much more likely to mean here 'true', as used of propositions. Moreover, Bluck's interpretation leaves the words both about a reason and about everything else' (a6) inexplicable. For they strongly suggest that the method was used in other spheres besides the inquiry into 'reasons'. Yet the idea of 'putting down as genuine things that conform to a definition' has no wider application. By contrast, a propositional interpretation allows us to see the hypothetical method in operation elsewhere. See on 84d4—85dl0,92cll-e3,93e7-94b3 (p. 166).

(3) The immediate context is, however, all-important. For it is to be expected that Socrates' account of his procedure should be adapted primarily to the matter in hand. If so, the method will be best understood by attending to its role in the quest for 'reasons'. The ensuing passage (100b—101c) may best be regarded as illustrating not only 'the theory judged strongest', but also the putting down as true whatever things seem to accord with it, and as not true whatever do not. Applications of the hypothesis that F things are F because they participate in the Form F are 'in accord' with that theory, whereas alternative 'reasons' yield statements that are not. Thus, 'x is beautiful because it participates in the Form Beautiful' and 'x is large because it participates in the Form Large' may be put down as true, since they are 'in accord' with the relevant Form-Reason hypothesis; whereas 'x is beautiful because of its colour or shape' and 'x is larger by a head' are not 'in accord' with the relevant hypothesis, and may therefore be put down as not true. 'Accord' is not, on this view, used in a sense equivalent to bare logical consistency. Socrates does not mean to accept just any proposition that may be logically consistent with the Form-Reason hypothesis. But 'accord' seems a natural enough term to use for the relation between the Form- Reason hypothesis and its applications. If, as suggested in (1) above, the hypothesis includes not only the assertion that the Forms exist, but also the thesis that F things are F 'for no other reason' than that they participate in the Form F (100c5—6, d4—5), then clearly the rejected alternatives will, in the sense just indicated, be 'not in accord' with it. The great mass of propositions, having no relevance to the issue about which any given hypothesis is put forward, will be neither 'in accord' nor 'not in accord' with it. They simply lie outside the scope of Socrates' remarks altogether.

This interpretation has the advantage of making Socrates' remarks directly relevant to their context. It also enables the plurals at 100a4—7 (the 'things' that do or do not accord), and the words 'on each occasion' (a4) to look forward to the plurality of examples that Socrates will give at 100c—101c. Thus, the present lines may be taken as simply enunciating a general schema into which those examples can be fitted.

For a somewhat similar interpretation see P. Plass, op.cit. 104—5. See also on 101c9-102a9.

lOObl—c8. Socrates says that he will 'display' (1) the kind of reason with which he has been dealing (b3—4), and (2) 'the reason' (b8). He will thus (1) show what the general requirements for a 'reason' are (see next note and on 100e5—101b8), and also (2) indicate the particular 'reason' that he is seeking, i.e. the reason for coming-to-be and destruction. The word 'display' (cf.99d2) suggests that he will exhibit these things in what he says rather than state them directly.

Once again the Theory of Forms is readily accepted (cf.65d6, 74b 1). At 100b5—6 the literal meaning could be 'something is beautiful alone by itself', or 'there is something beautiful alone by itself', or 'beautiful alone by itself is something'. See on 74a9—bl and note 5. The translation is based on the third interpretation. Cf.l02bl.

The next step (c3—8) is to agree that things are beautiful for no reason save that they participate in the Form Beautiful. R. Robinson (P.E.D. 127) takes 'what comes next to those things' (c3) to mean 'what logically follows'. But the present step is not so much a logical consequence of the hypothesis that the Form Beautiful exists as an integral element in it. 'Next' may mean only that it belongs next in an orderly statement of the argument—cf. Gorgias 454c 1-2.

'If anything else is beautiful besides the beautiful itself' (c4—5): note that the words 'besides the beautiful itself' clearly imply that the Form Beautiful is beautiful. This raises the question whether the character of beauty is being attributed to the Form, or whether 'the Form F is F' should be understood in some other way. See on 74d4—8. For Beauty the 'self-predicative' interpretation is defensible (see G. Vlastos, R.M. 1972, 456). But what of the Forms Numer- ousness and Twoness (101b6, 101c5)? Could Twoness be two, or Numerousness numerous, without wrecking the base of the Theory, that there is just one Form for each set of things to which we apply a common name (.Republic 596a, 597c)?

The nature of the relation between beautiful things and the Form Beautiful will be deliberately put aside at 100d5—7 (see next note). The word translated 'participate' (c5, 101c3-6) is the ordinary Greek word for 'share' used semi-technically. 'Partake' is used similarly at 102b2. To say that beautiful things 'share' Beauty is to say that they have that feature in common. The relevant sense of 'share' is that in which x and y may share A, without its being the case that each of them has only a part of A (as two people may share an ancestor or a birthday). This point is exploited in the Parmenides (131a—e), where 'share' and 'partake' are wilfully mis­construed.

100c9—e4. The Form 'reason' for things being beautiful is now further elaborated. For the text at 100d5—7 see note 63. Socrates remains non-committal as to the relation between Forms and particulars. This relation is the focal point of criticism of the Theory in the Parmenides (131—5). See on 65d4—e5 and previous note. The language of 'presence' was perhaps already a source of sophisti­cal objection. Cf. Euthydemus 300e—301b.

It is obvious why such 'wise reasons' as colour and shape are to be rejected. No given colour or shape is either a necessary or a sufficient condition for a thing's being beautiful. The reason why the Form answer would be 'safest' (d8, el, cf.l01d2, 105b7) is that any other answer could be refuted with counter-examples: a certain colour or shape might be present in a thing, and yet that thing might not be beautiful, or might be ugly; and other things might be beautiful, even when that colour or shape was lacking. But the Form reason, and it alone, would be 'safe' from all such objections: participation in the Form F is both a necessary and a sufficient con­dition for a thing's being F. The 'safety' in question is immunity from rebuttal by the counter-arguments sketched at 97a5—b3 and 101a5—b2.

Forms, interpreted as 'reasons', should not be taken as Aristot­elian 'efficient causes'. The Form Beautiful, for example, should not be taken as a beautifying agent, which is somehow supposed to impart beauty to things, or generate beautiful objects. For this interpretation, which naturally thrives on the mistranslation of aitia as 'cause', see Aristotle, De Gen. et Corr. 335b7—16, and Hackforth, 144—5. If the Forms are causes, Aristotle asks, why do they not generate things continuously instead of intermittently? Hackforth thinks that Aristotle here fastens upon 'the weakest point in the theory', viz. its failure to explain what causes the acquisition of attributes. But since the Forms are not represented as explaining that, the criticism is irrelevant. See G. Vlastos, P.R. 1969, 303-7, E. L. Burge, Phronesis 1971,2, n.4.

On the other hand, the text should not be over-interpreted in another direction. Apart from exhibiting the requirement that the 'reason' should be a necessary and sufficient condition for any given concept, and insisting that Forms exist, the Form-Reason hypothesis is wholly uninformative. This, no doubt, is why it is called 'simple- minded' (d4) and 'ignorant' (105cl). It gives no analysis for any of the concepts mentioned. The appeal to Forms side-steps rather than performs this task. Vlastos (op.cit. 314—15) formulates plausible logico-mathematical conditions that such an analysis might yield, in the case of 'numerous' (101b4—6), and represents these conditions as 'what Socrates is telling us, put into more modern language'. However, Socrates refrains from 'telling us' any such things. All we can say is that participation in a Form requires, in each case, that some such conditions be satisfied. We are given no insight, in any particular case, into what they are.

Moreover, it need not be supposed that Socrates' sole concern in the wider argument is with the analysis of concepts, or the formulation of necessary and sufficient conditions. Even when it is recognized that a thing (x) is F in virtue of its participating in the Form F, and even when this latter notion is properly understood, there remains the question how x 'comes to' participate in that Form. Socrates does not answer this question here. But he can plausibly be seen as doing so at 105b—d. So the denial that Forms are 'efficient causes' does not entail that an interest in such causes is altogether extraneous to the argument. See on 105b5—c8 (p.211).

100e5-101b8. The Form-Reason hypothesis is now applied to the cases of 'large' and 'numerous'. Note that at 100e5 the Form Largeness is given as the reason not only for large things being large, but also for larger things being larger. Separate Forms are not posited for comparative adjectives. It is the Form F that accounts for things' being 'more F', just as it does for their being simply 'F'. Cf. also 101b4—7. In general, 'F' and 'more F' are subjects of the same conceptual inquiry. This would explain some looseness earlier (75c9), where Socrates spoke of 'the larger' and 'the smaller'. See on 70e4—71all, 75c7—d6.

G. Vlastos calls this feature of the account a 'blemish'. It is, he thinks, one of Plato's 'residual confusions and fallacies' that he has failed to bring out that the instances under discussion are 'special cases of the "greater than" relation, and that the absolute numer- ousness or bigness of the things he is talking about is irrelevant to the reasoning' (P.R. 1969, 315, n.64). But this overlooks the role of the Forms in connection with such concepts as 'large' and 'numerous'. It is just because particular things or groups are never 'absolutely' large or numerous, that Forms for those concepts are introduced. The connection between comparative adjectives and Forms is made explicit at 102b—c, where Socrates explicates 'larger' and 'smaller' with reference to the Forms Large and Small. This makes it highly unlikely that Plato has ignored, let alone overlooked, the difference between comparative and simple adjectives. See on 102al0—d4.

At 101a5-b2 and 101b4-7 Socrates expresses 'fears' that throw light on what was wrong with the reason rejected earlier (96d8-el) for one man or horse being larger than another. There would be two things to fear in giving 'a head' as the reason.

(1) 'If you say that someone is larger and smaller by a head, then, first, the larger will be larger and the smaller smaller by the same thing' (101a6—8). The apodosis of this sentence could mean either

x is larger than y by a head and y is. smaller than * by a head, or

x is larger than y by a head and smaller than z by a head. With (i), the phrases 'the larger' and 'the smaller' will refer to two different items, x and y, with the converse relations of 'larger than' and 'smaller than' holding between them. On this view, the supposed 'contradiction' could be formally deduced from ^lx is larger than y by a head', given that:

(x, y) {(x >y by h) = (y < x by h)}. With (ii) 'the larger' and 'the smaller' will refer to the same item, x. which will be 'larger' and 'smaller' in relation to two different things, y and z. On this view, the 'contradiction' could not be formally deduced, since from (x >y by h) we cannot derive (3z) (x < z by h). But (ii) seems to follow more naturally from the antecedent 'if you say that someone is larger and smaller by a head', which mentions only a single subject, (ii) would seem a more typically Platonic way of making the point that particulars may have contrasted predicates in different relations. And it has the advantage of enabling the later discussion of 'Simmias is larger than Socrates and smaller than Phaedo' (102b—d) to be linked directly with the present example.

On either interpretation the essential point is that 'a head' could just as well be viewed as a reason for 'being smaller' as for 'being larger', and therefore cannot be what constitutes anything's being larger. Hence it cannot help to answer the conceptual question 'what is largerV Evidently, it is here being assumed that if any feature is constitutive of a characteristic F, that feature will be found in all and only those things that are F, and not in things that are G. These lines thus fulfil the promise of 100b3—4 to 'display' the sort of 'reason' with which Socrates is concerned. For they exhibit, without expressly stating, the requirement -lat the 'reason' in question should be both a necessary and sufficient condition of the concept that has to be explicated. Whatever is the 'reason' for a thing's being F, or more F, cannot also be the reason either for its or for anything else's being G, or more G.

Vlastos (loc.cit.) assumes interpretation (i), and objects that the fact that y is smaller than x by a head would be a spurious reason for rejecting 'a head' as what makes x larger. For, he urges, x and y are different items, and 'there is no contradiction in the same cause producing contrary effects on different things'. On interpretation (ii) this difficulty does not arise, since only a single item, x, will be involved. But even on interpretation (i) Vlastos' objection does not affect Socrates' point. For he is, in effect, here stipulating what shall count as a 'reason' for any given property, and he is excluding as a reason for F anything that features in cases of F and G alike. This point can be made whether the items concerned are the same or not.

See on 96e6-97b7.

(2) The other objection to 'a head' as the reason for one man's being larger than another is (a8-b2) that a head is itself a small thing, and 'it's surely monstrous that anyone should be large by something small'. This displays a further requirement for a 'reason'. Whatever is to be a reason for x's being F must not itself be characterized by F's opposite, G. No adequate 'reason' for a property, that is genuinely constitutive of it, can possess the opposite of that property. So a head, being characterized by smallness, cannot be what con­stitutes a thing's being larger. The examples at 101b4—7 can be understood similarly. Both the rejected answers, 'two' and 'half of two cubits', are to be thought of as something 'small', and therefore ineligible as 'reasons' for anything's being 'large', or 'larger'.

Here again Vlastos (op.cit. 316, n.64) finds the argument flawed. The fact that a head is a small thing would not, he objects, preclude it from making x larger, as distinct from large. For* may be a larger man than y, without being a large man. But this objection seems, once again, to miss Socrates' point. The difficulty is: how can something'small'or (in this case) 'smaller than y' be what constitutes x's being 'large' or (in this case) 'larger than y"> How could something 'small' or 'smaller' be the true 'reason' for anything's being 'large' or 'larger'? By calling this supposition 'monstrous' (bl), Socrates exhibits a further condition that a true 'reason' must meet.

His objections to the rejected reasons are here formulated in an eristical manner, no doubt in parody of the contradiction-mongers (101el-2). But the principles implicit in these objections have a serious role in the coming argument. Putting them together with the one noted earlier (see on 96e6—97b7) the requirements for a 'reason' may be summarized as follows:

No opposite, F, can count as the 'reason' for a thing's having a property, if its opposite, G, can also give rise to that property (97a7—b3).

Nothing can count as a 'reason' for a thing's having a property F, if it can also give rise to theoppositeproperty G(101a6—8).

A 'reason' for a thing's having a property F, cannot itself be characterized by the opposite of that property, G (101a8—b2).

Requirement (iii) is crucial for what follows. For when Socrates comes, later on, to improve upon the present 'safe* Form-Reason hypothesis with 'a different kind of safeness' (105b8), he does so with examples that are 'safe' from the objection here brought against 'a head'. Fire, fever, and oneness (or a unit) cannot be characterized by the opposites of the properties of which they are 'reasons'. Soul is a 'reason' of the same type. What qualifies it as a 'reason' for something's being alive is precisely what disqualifies 'a head' as a reason for something's being larger. Hence, in this trivial example a major principle underlying the final proof of immortality can be discerned.

The principle in (iii) is, however, highly questionable. It is well discussed by E. L. Burge, Phronesis 1971, 5. See also on 105b5—c8 (p.213), 105c9—dl2.

101b9—c9. The grounds for rejecting addition and division as reasons for things' coming to be two, or one, were given earlier (97a7—b3). The reason now recommended is that they participate in the Forms Twoness and Oneness. The assimilation of number concepts to others tends to mask the peculiar difficulties to which they give rise (see on 96e6—97b7). Adequate treatment of those problems would call for an inquiry into number concepts such as Plato envisages at Republic 525a—526c. But it is not to his purpose to pursue it here.

Plato uses two kinds of words for numbers: (1) the ordinary words for the series of cardinal numbers, and (2) a series of words ending in -as, (monas, duas, trias, etc.). It is uncertain whether any consistent distinction is intended between these, and in particular whether members of (2) refer exclusively to Forms, and whether members of (1) never do so. In the present passage members of (2) are used, and the Forms Oneness and Twoness are clearly meant. In the next few pages, however, the matter is often more debatable. For this issue, which becomes crucial for the interpretation of the whole argument, see J. Schiller, Phronesis 1967, 57—8, and D. O'Brien, C.Q. 1967, 217-19. The formal difference between the two series has been marked in translation by adding '-ness' for members of (2): 'oneness', 'twoness', 'threeness', etc., and omitting it for members of (1): 'one', 'two', 'three', etc. The English 'monad', 'dyad', and 'triad' are transliterations of the roots of words in series (2), but their special associations make them unsuitable as translations. The present renderings may, however, wrongly give the impression that the abstract character possessed by, e.g., sets of three things is meant. The Greek words contain no suggestion of this. The termination '-ness' has been used simply to mark the formal difference between these words and members of series (1). In this way no questions of interpretation are begged, or, of course, solved. See on 103e5-104c6, 104d5-e6, 104e7-105b4.

101c9—102a9. Socrates here resumes his comments on hypo­thetical method, interrupted since 100a. They are now framed as precepts to Cebes, but are presumably to be taken closely with what

Socrates has said that he does himself. At 101 d 1-2 he says you 'would hang on to that safety of the hypothesis, and answer accordingly', i.e. in the way specified at 101c2—7, by appealing to the relevant Form. The phrase translated 'that safety of the hypo­thesis' is somewhat awkward. P. Plass {Phronesis 1960, 111—12) would translate 'that safe consequent of the hypothesis', supplying 'consequent' from 101d4 below. But this seems neither natural nor necessary, if the 'hypothesis' is not merely that Forms exist, but that things are F by participating in the Form F, i.e. the Form- Reason hypothesis of 100b5—c8. This answer you would give, Socrates tells Cebes, being 'scared of your own shadow' (c9-dl). Here he alludes to the risk incurred by any departure from the Form-Reason hypothesis. Any other answer to the question 'what makes F things F? would be prey to rebuttals of the sort illustrated at 101 a-b.

At 101d3-5 he continues: 'But if anyone hung on to the hypothesis itself, you would dismiss him, and you wouldn't answer till you should have examined its consequences to see if, in your view, they are in accord or discord with each other'. It is far from obvious what is supposed to give rise to this examination, or what it consists in. In particular, its intended scope is unclear. Is Socrates giving general guidance on hypothetical method? Or is he concerned merely with its use in the quest for 'reasons'? Note that his directions arise out of the example at 101c. 'The hypothesis itself (d3) must refer back to the hypothesis whose safety is clung to at 101 d 1—2, viz. that things come to be (are) two (one) for no other reason than that they participate in the relevant Form. This link with the preceding passage suggests that the present lines should not be isolated from their context. Here as at 100a the hypothetical method needs to be understood, initially at least, in application to the Form-Reason hypothesis itself. For the meaning of 'hang on to' at lOldl and 101d3 see note 67.

What are the 'consequences' of the hypothesis? The words trans­lated 'consequences' (d4, e3) are not the usual technical term for logical conclusions, but are passive forms of the verb Socrates had used at 100a3 to mean 'proceed'. It seems, however, difficult to understand it here in anything but a logical sense. For unless some sort of test for propositional accord or discord is envisaged, no sense can be made of the examination to see if the things in question are 'in accord or discord with each other'. The word has therefore been translated 'consequences'. Cf. R. Robinson, P.E.D. 129—31, R. C. Cross, P.R. 1956, 406. For a different view, see Bluck, 14, 169.

What could be meant by asking whether the consequences that spring from the Form-Reason hypothesis are 'in accord or discord with each other'? It has seemed puzzling that Socrates should envisage testing the consequences of a hypothesis that he has already adopted as the 'strongest' (100a4), and with which, therefore, he should have been satisfied already. Moreover, if 'accord' and 'discord' here mean 'consistency' and 'inconsistency', how could consequences springing from a single hypothesis fail to be 'in accord' with each other? For no single proposition can logically entail consequences that do, in fact, contradict each other. For these difficulties, see Robinson, op.cit. 131—3. A hypothesis may, as he suggests, be viewed as having conflicting consequences if it leads to contradiction when combined with other standing assumptions. It was in just this way that Simmias' hypothesis that the soul is an attunement was disproved, (see on 92cl l-e3,93e7-94b3, p.166).

Could the Form-Reason hypothesis, which Socrates has insisted is 'safe' (lOOel), be thought to give rise to contradictions of that sort? Its 'consequences', e.g. that x is large because it participates in the Form Large, and that y is large for the same reason, and likewise z, would necessarily all 'accord' with each other, and could present no contradiction. How, then, could the question whether the con­sequences of that hypothesis were in accord or discord with each other possibly arise?

A possible answer is that Socrates here has in mind sophistical opponents of the Theory of Forms. The Forms, they might object, are no less vulnerable than particulars to the 'contradictions' developed at 101a—b. An assault of that sort is mounted against the Theory in the Parmenides, where 'monstrosities' are proved with respect to the Forms Large and Small (13 la—e), that are recognizably similar to the 'monstrosities' (101b 1) that the Theory is here designed to avoid. The present passage need not, of course, be held to refer to the particular arguments given in the Parmenides. But it seems natural that Socrates should tell Cebes that he should be ready to face arguments of this general type. For it would be of special importance, for anyone whose faith in rational argument rested upon the Theory (cf.90c9—dl), that it should be immune from the diffi­culties that gave rise to it. The fact that Socrates ends his present remarks with a slighting allusion to 'contradiction-mongers' (e2) suggests that a supporter of the Form-Reason hypothesis would, indeed, have to defend it against them.

In this way, perhaps, 101c9—d5 can be related to the hypothesis at 101c that was its point of departure. But now (d5— el) Socrates moves to a new and still more difficult stage: 'and when you had to give an account of the hypothesis itself, you would give it in the same way, once again hypothesizing another hypothesis, whichever should seem best of those above, till you came to something adequate'.

What is meant here by 'giving an account' of the hypothesis itself? For different meanings of 'give an account', see on 76b4—clO. The most natural sense for it to bear here is 'give a proof', i.e. support or justify the initial hypothesis. Socrates seems to be thinking, as Robinson says (op.cit. 136), of the objector who says 'Yes, your conclusion follows from your hypothesis; but how do you know the hypothesis is true?' This new challenge is understandable. For it would not, presumably, be sufficient to adopt a hypothesis ad hoc, merely because it would yield one's desired conclusion. There must be some further ground that would recommend it independently. This ground is to be given 'in the same way', i.e. in the same way as the initial hypothesis was put forward. It, in its turn, will be related to a further hypothesis, in the same way as the original conclusion was related to it.

But here a problem arises. Socrates' remarks seem by now to have cut loose from their moorings at 101c. For the original 'safe' hypothesis, to which reference has so far been made, was that F things are F for no other reason than that they participate in the Form F. Yet the relation between this and the things it was intro­duced to explain is unlike the relation between premiss and con­clusion. How, then, could a hypothesis designed to justify the Form-Reason hypothesis be posited 'in the same way' as was that hypothesis itself? If, as Robinson plausibly maintains (op.cit. 137), the words 'whichever should seem best of those above' (d7) are taken to mean that the relation between successive hypotheses is one of entailment, then the Form-Reason hypothesis would have to follow from 'another hypothesis' in the same way as its entailments follow from it. Yet it seems hard to find in the text, or to supply, any 'higher' hypothesis to which the Form-Reason hypothesis is thus related.

There is further obscurity in the clause 'till you came to something adequate' (el). Does this mean merely 'adequate to satisfy an objector to the first hypothesis'? Or 'adequate to satisfy yourself'? Robinson (op.cit. 137) excludes the latter, on the ground that 'you were already satisfied with the first hypothesis'. But this seems doubtful. For in dialectic the true philosopher will be his own objector. However strong his hypothesis may seem to him, it behoves him to justify it not only to his interlocutors but also to himself. Socrates has to persuade himself as well as others (91a6—bl, cf.lOOel—2).

It should also be asked whether 'something adequate' means 'some adequate hypothesis', as Robinson supposes, or whether it will consist in, something that is no longer hypothetical in character. A mere hypothesis, it might seem, could not be 'adequate', if its sole merit were that no objection to it could be found. A conclusion, however validly derived from a hypothesis, would be no stronger than the hypothesis itself; and if there were no positive reason to adopt the latter, it would afford no adequate ground for the con­clusion. The Theory of Forms is a case in point. It will not give adequate ground for believing in immortality, unless there are independent grounds for adopting it. This, no doubt, is what Socrates means later (107b5—8), when he tells his listeners that they must study 'the initial hypotheses' further, even if they find them accept­able, and that they will follow the argument if they analyse them 'adequately'. See on 65d4-e5 (p.97), 107a2-bl0.

What might such an analysis consist in? The words 'best of those above' (d7) are sometimes taken to suggest an ascent, via successively 'higher' hypotheses, to a 'starting-point' (cf.e2), i.e. to some ultimate certainty that is not itself a hypothesis, but from which the pro­positions so far hypothesized can be deduced. Such a starting-point would be 'adequate' in the sense that it needed no justification itself, and rendered the system of propositions derived from it not only logically coherent but also true. In the Republic (510b7, 511b6) a starting-point of this kind is called 'unhypothetical', and is identi­fied with the Form of the Good. Since Socrates' present account of his method was preceded by the story of his fruitless search for 'the good' (99c6—8), the present move from hypotheses to 'something adequate' has often been thought to anticipate the ascent to a first principle in the Republic. Accordingly, some commentators have wished to understand these lines in terms of a hierarchy of tele- ological propositions, somehow culminating in the Good. See, e.g., R. S. Bluck, Phronesis 1957, 21-31. The text, however, gives this no explicit support. No doubt Plato envisaged a system of ordered hypotheses as an ideal, for methodical scientific inquiry and exposition. But we can barely conjecture how, in detail, this ambitious programme was to be carried out. See, further, P. Friedlander, C.P. 1945, 256, H. F. Cherniss, A.J.P. 1947, 141, M. D. C. Tait, S.H.G.N. 110-15.

Socrates ends (lOlel—102al) by telling Cebes that he would not mix things up, like the 'contradiction-mongers', by arguing at the same time both about the starting-point and its consequences. The broad sense of this seems clear: the starting-point of an argument should be examined separately from the propositions derived from it. Aristotle attributes a similar precept to Plato (E.N. 1095a32), and it is well suited to the analysis of Platonic arguments generally. There may, however, be a more specific allusion to the contradiction- mongers' technique of confusing propositions about Forms with propositions about the particulars named after them. Such principles as 'it is by the beautiful that beautiful things are beautiful' (100e2— 3) would be open to misconstruction, in view of the ambiguity of 'the beautiful'. Just such a confusion in the meaning of 'the larger comes to be from the smaller', or more generally 'opposites come to be from their opposites', occurs later in the discussion (103a4— c2), and has to be disentangled. Cf. Euthydemus 300e—304b, where such ambiguities are exploited, and the exploiters castigated in terms similar to those used here—cf,101e5—6 with Euthydemus 303c—d. See also on 90b4—91c5. For the phrase 'to discover any of the things that are' (e3) see on 65c2—4.

3. 7 The Final Argument (102al0~-107bl0) Socrates now advances his final proof that soul is immortal and imperishable. Soul, which brings life to the body, cannot admit death, and is therefore immortal. And since the immortal is imper­ishable, soul cannot perish, but must withdraw at the onset of death.

102al0-d4. Socrates explicates the statement 'Simmias is larger than Socrates but smaller than Phaedo' in terms of the Theory of Forms. 'Large' and 'small' have been used in translation, rather than 'tall' and 'short'. But since the three men are, presumably, being compared in respect of height, 'overtop' has been used for the verb that expresses the relation between them.

For 'each of the forms was something' (bl) as a way of asserting their existence, see on 74a9—bl. At 102b2 the particulars that 'partake' in a Form are said to 'take its name'. For the Forms as 'eponymous' see on 65d4—e5 (p.96), 78dl0-e5. Individuals, such as Simmias, are said to 'take the name' of large and small (clO—11). Cf.l03b7-8. Note that whether a thing is called 'large' or 'larger', it is regarded as named after one and the same Form, the Form Large. Cf.l00e5—6, and see on 100e5-101b8.

Part of Socrates' purpose, evidently, is to distinguish properties that belong to a subject by its nature from those that it merely happens to have, and that it could lack without ceasing to be itself. Simmias does not overtop Socrates 'by nature' (cl), in the way that three and five will later be said to be odd 'by nature' (104a3, a7). He does not overtop Socrates 'by virtue of his being Simmias', but by virtue of the Largeness that he 'happens' to have (c2). This language marks the contrast between what we should call 'essential' and 'accidental' predication, which will be of the first importance in the coming argument.

Beyond this, however, the analysis of comparative statements is problematical.

(1) Why should theTstatement 'Simmias is larger than Socrates' be held not to be strictly or literally true (b8—cl), in virtue of its ascribing to Simmias an accidental property? Is it implied that all statements of accidental predication, including non-relational ones, need reformulation? And is it further implied that no statements of essential predication, not even relational ones, need such recasting, so that not only 'three is odd' but 'three is greater than two' is acceptable as it stands? If this is being suggested, the grounds for requiring a reformulation in one sort of case but not in the other are not explained. But it seems possible, despite the suggestion of 102b8—cl, that the need for reformulation arises more from the relational than from the accidental feature of the example. For it is relational predicates, whether they be accidental or essential, that give rise to the compresence of opposites in a single subject: 'three is greater than two but less than four' is just as true as 'Simmias is larger than Socrates but smaller than Phaedo'. Has Socrates conflated what we should regard as two different distinctions?

O'Brien (C.Q. 1967, 200, n.l) says that the relational feature of the example is incidental: 'the colour of an apple, which is an accidental but not a relative attribute, would have served equally well as a contrast to, and as a preparation for, the essential hotness of fire and the essential aliveness of soul'. But the case of an apple's colour would not be a foil to the principle to be maintained at 102d5—103c9, that opposites exclude opposites. For an apple cannot be red and green all over at the same time, as Simmias can be larger than Socrates and smaller than Phaedo at the same time. Yet it is the compresence of opposites in a particular that seems signifi­cant. For it will next be argued that Largeness, unlike Simmias, cannot be large and small 'at the same time' (102d7, cf.e8). See next note.

(2) What is gained by interpreting 'Simmias is larger than Socrates but smaller than Phaedo' in terms of Largeness and Smallness being in Simmias (102b5—6)? It may be intended to clarify the notion of largeness, the simple term 'large' being held to be covertly comparative. To say that an individual is large is to say that it is a large member of some class, and therefore larger than most, or than average, members of that class. If so, an account of 'large' will require an account of 'larger'. Yet this will seem unsatisfying, if one expects the simple term to be primary, and the comparative to be explicable in terms of it. The proposed analysis of the comparative statement as 'largeness and smallness are in Simmias' (b5—6) may, therefore, be an attempt to restore primacy to the simple adjective by treating Largeness and Smallness as 'relational properties'. Cf. I. M. Crombie, E.P.D. ii. 312.

But the analysis gives rise to a difficulty. (SI) 'Simmias is larger than Socrates' means, we are to understand, (S2) 'Simmias has Largeness in relation, to Socrates' Smallness.' And this, apparently, means (S3) 'Simmias' Largeness overtops Socrates' Smallness' (102dl— 2). But since 'overtops' is equivalent to 'is larger than', S3 will imply (S4) 'Simmias' Largeness is larger than Socrates' Smallness.' But if so, how is the ascription of 'larger than' to the Form Large­ness in S4 to be understood? It looks as if it is 'self-predicative' in the way that was earlier seen to give rise to paradox. Thus, given that:

(x, ,y){(x is larger than y) = (x has Largeness in relation to y's

Smallness)},

then, where x and y are Simmias' Largeness and Socrates' Smallness respectively, S4 will yield (S5) 'Simmias' Largeness has Largeness in relation to Socrates' Smallness.' Here a regress threatens, akin to that of Parmenides 132a—b. If, on the other hand, S4 is not self- predicative, it is not clear what it should be taken to mean. It does not seem readily construed as an identity statement, in the manner sometimes proposed for 'the Form Equal is equal' (see on 74d4-8, p.128). How else could the predicate 'larger than' or 'overtops', as applied to the Form Largeness, be understood?

102d5— 103a3. It is now argued that opposite Forms cannot be characterized by their opposites. For the translation at 102e5—7, see note 69.

Several problems arise here:

The connection of thought expressed by 'I want you to think as I do' (d5) is not obvious. Why should the foregoing analysis of comparative statements have been given in order that Cebes may share Socrates' view on this new point? Perhaps the analysis was meant to dispose of cases in prima facie conflict with the principle being maintained here. The principle that opposites exclude each other might seem to be breached by particulars in which Largeness and Smallness can coexist. But, on the analysis just given, the principle is not really infringed. For the analysis at 102cl0—d2, even if it requires 'larger' to be predicated of Largeness, and 'smaller' of Smallness (see previous note), does not require Largeness and Small­ness to be ascribed to each other. The question of the relation between Simmias' Largeness and that of Phaedo, or between Simmias' Smallness and that of Socrates, would, no doubt, create embarrassment, given that:

(x, y) {(x is larger than y) = (y is smaller than x)}. But the analysis adroitly avoids asserting any relation between the Largenesses or between the Smallnesses of two individuals. The principle that opposites exclude each other therefore remains intact.

Socrates expresses the relations between Forms and their opposites in a series of 'military metaphors'—'advancing', 'getting out of the way' (i,e. retreat), 'abiding' and 'admitting' hostile forces that 'occupy' a garrison, and 'perishing'. These metaphors will be kept up right through to 106e7. How are they to be applied to Largeness and Smallness? In particular, are 'getting out of the way' and 'perishing' genuine alternatives in this case, and if so, how do they differ?

Hackforth (148, n.3) believes that 'getting out of the way' is brought in here only to provide for the case of soul, the single case in which it is exemplified. This would suggest that the alternative is not meant as a real one in other cases. But in view of the repeated stress upon it, and its wide range of application (e7—a2), it seems better to find distinct interpretations for each option, if possible (see on 103cl0-e5, p. 198). It is plausible to suppose, with D. O'Brien (C.Q. 1967, 204, n.4), that Smallness 'advances' when Simmias is compared with Phaedo, and that his Largeness 'gets out of the way' precisely because, despite the comparison with Phaedo, he remains larger than Socrates. If this is correct, it will be prefer­able to think of 'perishing' as needing something stronger than mere cessation of the comparison with Socrates. For if Simmias' Largeness in relation to Socrates does not depend upon their actually being compared, it need not 'perish' when the comparison ceases. For 'perishing', then, an actual change of size in one of them will be required. Cf, Theaetetus 155b6—cl.

(3) The Forms Large and Small are pointedly distinguished from what Socrates calls 'the largeness in us' (d7) and 'the small that's in us' (e6). The contrast is repeated at 103b5, and must be significant. What is meant by 'the F in us', and what is its role in the argument?

Whether Plato distinguished a category of 'immanent Forms', with separate ontological status from the Forms proper, is disputed. The distinction is accepted by Hackforth passim, Bluck, 17—18, R. Demos, P.P.R. 1947-8, 456-60, R. G. Turnbull, P.Q. 1958, 131-43, and G. Vlastos, P.R. 1969, 298-301. It is denied by Verdenius (note on 103b5), and by D. O'Brien, C.Q. 1967,201-3.

The phrase 'immanent Form' suggests a more systematic doctrine than the evidence warrant?. Certainly, no consistent distinction between 'immanent' and 'transcendent' Forms can be founded upon Plato's terminology (see note 72). He may, more aptly, be said to distinguish here the property Largeness from individual instances of it. Such a distinction arises naturally out of the preceding talk of Simmias' and Phaedo's Largeness (102c4, 102cll). A somewhat similar notion, for the Form Likeness, appears at Parmenides 130b3— 4. But the scope of the distinction between Forms and their instances 'in us' remains unclear. Is it only for Forms of attributes, such as

Largeness and Smallness, that property-instances are distinguished? Or are there Forms 'in us' for all items for which Forms may be postulated, including such stuffs as fire or snow? Moreover, for those Forms that are explicitly distinguished from their property- instances, what counts as an 'instance' of the Form in question? Is it Simmias himself that instantiates the Form Largeness, or is it only 'the largeness in him' that does so? Or may both be thought of as instantiating the Form in different ways?

Doubt on these points leaves it uncertain what part, if any, 'the largeness in us' plays in the argument. Since the alternatives 'get out of the way' and 'perish' are applied to it, as they will later be to snow and fire, three, and soul, 'the largeness in us' has sometimes been regarded as their precursor in the argument, and they have been construed accordingly as 'immanent Forms' (see on 103cl0— e5). But it is not clear that 'the largeness in us' is meant as analogous to the disputed items in the later argument. Note that those items have no counterparts related to Large and Small in the way that they themselves are related to the Forms of Hot and Cold, Oddness, and Life. In terms of Socrates' 'safe' and 'subtle' reasons (105b—c), the only 'reasons' given for things being large or small are 'safe' ones. There is no entity which 'makes' things large or small, in the way that fire or snow, three or soul, (however interpreted), 'makes' them hot or cold, odd or alive. It is far from certain, therefore, that interpretation of the later argument should be controlled by the present distinction.

A possible way of relating the distinction to the argument will, however, be considered below. For this purpose it will be necessary to distinguish between the property-instances of a Form (e.g., Simmias' Largeness) and the individual things that may be thought of as participating in that Form (e.g. Simmias himself). Property- instances, since their designation must include the name of the Form in question, will be referred to in what follows as its N- instances. The individual things participating in the Form will be called its T-instances. Thus, 'the largeness in Simmias' will be said to be an iV-instance of the Form Large, whereas Simmias himself will be called a ^-instance of it. For the application of this distinction see Version B given in the notes on 104c7—d4 (p.204) and 104d5- e6 (p.207).

103a4—c9. An objection is now raised-against the principle that an opposite F will never come to be G: it appears to conflict with the law of opposites agreed earlier (70el—71all). Socrates shows that the objection rests on a misunderstanding, thereby revealing an ambiguity in such expressions as 'the large': they can mean either 'that which is large' or 'the property of largeness' (see on 70e4—71all, p.108). In the former sense 'the large' can come to be small, in the latter it cannot. Despite this clarification, however, expressions of the form 'the F' will continue to give trouble, notably at 106a-e. See on 105e 10-107a 1 (p.217).

To draw the distinction between largeness and large things, Socrates has to emphasize the word 'thing' (103b3), that he had used more casually earlier (71al0, b2). It is sometimes remarked, in this connection, that 'things' and their 'attributes' had not, before Plato, been differentiated. Cf., e.g., the use of 'hot', 'cold', 'dry', and 'wet', at 86b8—9. Only later will Plato himself coin a word for 'quality' (Theaetetus 182a8). It should be noted that Greek commonly uses unaccompanied plural adjectives, e.g. 'beauti- fuls', 'larges', 'smalls', to mean 'F things'. Of the adjectives promin­ent in the Phaedo only 'equal' is naturally so used in English.

At 103b5 Socrates distinguishes 'the opposite in us' from 'the Opposite in nature'. By 'the opposite in nature' is meant the Form F itself, as distinct from its property-instances 'in us'. For this use of 'nature' in the designation of Forms, cf. Republic 597b-598a, Parmenides 132d2, Cratylus 389a—d. The language at 103b7—cl again reflects the theory that particulars are 'named after' an eponymous Form. See on 65d4-e5 (p.96), 78dl0-e5, 102al0-d4.

At 103a7—8 the law of opposites is recalled in terms of the com­paratives 'larger' and 'smaller'. Cf.70e6—71a2. For the shift from 'large' and 'small' at 102d5-103a2 to 'larger' and 'smaller' here, see on 70e4—71all, 100e5-101b8, 102al0-d4.

103cl0—e5. A new phase of the proof begins here. Note the semantic theory underlying the introduction of hot and cold into the argument (clO—11). Socrates asks, literally, 'do you call some­thing hot, and again cold?', i.e. 'is there something designated by each of those names?' At 103e3—4 the Form is said to be 'entitled to its own name for all time'. Once again, the Form F is treated as the prime bearer of the name 'F'. Cf.l03b7—cl, and see previous note.

At 103cl3—d4 Socrates distinguishes the Hot and the Cold from fire and snow. Are Forms of Fire and Snow meant here, or physical stuffs? Interpretation of the argument hinges largely upon this much vexed point.

The. language is non-committal. Socrates uses no Form-referring expressions in connection with 'fire' or 'snow' at any point. It is true that Forms, regarded as bearers of common names, i.e. 'universals' (see on 65d4—e5, p.96), should include Fire and Snow no less than Hot and Cold. A Form of Fire is explicitly mentioned in the Timaeus (51b8). Cf. also Parmenides 130cl—4. But granted the need for such Forms in Plato's scheme of things, it is a further question whether they are being referred to here.

G. Vlastos (P.R. 1969, 318, n.70) and D. Keyt (Phronesis 1963, 168, n.2) have wished to understand snow and fire as 'immanent Forms', parallel to the Form of Three at 104d5—7, and have inter­preted the whole argument accordingly in terms of entailment relations between Forms. Similarly, Hackforth (156, 162, n.3) thinks that the argument wavers between treating fire as a Form, and taking it as a concrete substance co-ordinate with snow. However, (i) one would expect fire and snow to be co-ordinate throughout, (ii) It seems hard to believe that at 106a3—10 either fire or snow could be anything but physical stuffs. For it could hardly be said of the Form of Snow, not even of an 'immanent' Form of it, that it would go away 'intact and unmelted' (106a5), or of the Form of Fire that it would not be 'put out' (106a9). Nor (iii) can fire be an immanent Form at 106b6, where Socrates speaks of 'the hotnessin the fire', unless it be supposed that there can be immanent Forms within immanent Forms. Above all, (iv) on the Vlastos-Keyt view, it seems impossible to interpret 'getting out of the way' and 'perishing' as genuine alternatives. Clearly, Socrates could not be suggesting that the Forms of Snow or Fire might 'perish', since all Forms are imperishable. But if it is 'immanent' Forms, parallel to the Large and Small 'in us' (102d7, e6), that are held to 'get out of the way or perish', how can these alternatives be distinguished in such cases as Snow and Fire? In those cases, it would not be possible, as it was with the Large and Small, to think of 'immanent Forms' as getting out of the way, when the relevant particulars were viewed in a different relation. For 'snow' and 'fire' are non­relational terms. But the alternative is presented as if it were a real one. The contrast that Socrates will draw at 106a—c between snow, fire, and three, on the one hand, and soul on the other, suggests that only in the case of soul is 'getting out of the way' a forced option. In the other cases, both options are expressly left open. Cf. D. O'Brien, C.Q. 1967,204,208.

On the other hand, if fire and snow are not Forms, but physical stuffs, there need be no difficulty in understanding 'get out of the way' and 'perish' as genuine alternatives: snow can melt, and fire can be put out, or they can be moved. Their 'perishing' by being melted or quenched is expressly suggested at 106a3—10, and the idea of their 'getting out of the way' by being moved, although nowhere mentioned, is not hard to supply. Burnet therefore seems correct in saying (note on 104dl): 'it has not been suggested that fire and snow are Forms, and it seems improbable that they are so

regarded.'

At 103e2—5 Socrates generalizes from the snow and fire examples: certain things, although not identical with the Form F itself, always possess that Form's character. For the words trans­lated 'form' and 'character' see on 65d4—e5 (p.93) and note 72.

What is meant by qualifying the generalization with the words 'in some cases of this kind' (e2)? Is it that (i) only for some, but not for all, of the opposite Forms mentioned, there are things other than those Forms that are always entitled to their name? On this view, it will be implied that for certain Forms, such as Large and Small, there is nothing corresponding to snow and fire, i.e. nothing that is always large, or always small, in the way that snow is always cold, and fire always hot. Alternatively, does Socrates mean that (ii) only some, but not all, stuffs are invariably characterized by one member of a pair of opposites? It will then be implied that certain stuffs, unlike fire and snow, may be characterized by either of a pair of opposites (at different times), and are not invariably characterized by one member of a pair only. _

'Some cases of this kind' is more naturally taken, as in (ii), to mean 'such things as fire and snow', than as meaning 'such Forms as Hot and Cold'. But what, on this view, would be ruled out by the reservation? O'Brien (op.cit. 211, n.2) suggests water, which may be either hot or cold, unlike fire, which can only be hot. But water is, arguably, always wet and never dry, and thus stands in the same relation to the Forms of Wet and Dry as does fire to the Forms of Hot and Cold. If the reservation is meant to exclude things that are not invariably characterized by any opposite, what would illustrate it? This question calls for further clarification of the concept of an 'opposite'. On a broad interpretation, it might be difficult to find examples of things that are not invariably characterized by some opposite or other. But the reservation would be pointless if its effect were to exclude an empty class.

103e5—104c6. A numerical example is now introduced. Three, while not itself an opposite, must always be characterized by one member of a pair, the Odd, and so must exclude the other, the Even. The argument is sometimes criticized here for assimilating numerical to physical examples. Thus, Hackforth (157) objects that snow's refusal to admit cold is 'a physical fact known through sense perception', whereas the corresponding truths about three and soul are 'statements about the implications of terms'. G. Vlastos (P.R. 1969, 321) speaks of 'Plato's reduction of physical to logical necessity in the Phaedo'. See also E. L. Burge, Phronesis 1971,11— 13. But the passage need not be read as embodying any doctrine about the nature of physical necessity. The fire, snow, and number examples may be taken as entirely subservient to the proof of immortality. For this proof, the vital point is not so much the assimilation of physical to logical necessity as the basing of metaphysical conclusions upon conceptual argument, the derivation of existential propositions about the soul from consideration of its essential nature. See on 105el0-107al (p.217).

It is true, however, that the numerical examples serve to bridge, and even conceal, a serious gap between soul and the physical stuffs with which it has so far been compared. Soul will be thought of as causally affecting bodies (105c—d), as imparting life to the bodies it occupies, just as fire and snow impart heat and cold. But this analogy is weakened by the fact that soul, unlike fire and snow, is not observ­able (79b7—15). Fire and snow can be recognized and observed independently of the bodies they occupy. Our knowledge that a body has fire or snow in it does not depend upon our finding it to be hot or cold. By contrast, the presence of soul in a body is not observable independently of that body's being alive. The life that we find in a body is our sole warrant for ascribing soul to it at all. Hence the notion of soul's 'bringing life' to body is not truly parallel with the other cases. The transition to it is helped by the example of three. For numbers, like the soul, are not sensible, nor are they observable independently of particular numbered sets.

The status of numbers in Plato is controversial. It is not clear whether he distinguished between the number n and the Form iV-ness. But the following considerations suggest the need for such a distinction: (a) When we speak of, e.g., 'four threes', or 'adding three and three', the threes mentioned can hardly be the Form Threeness, which is unique, (b) Forms are held to be 'incomposite' (78c), whereas numbers might be regarded as consisting of abstract units, and hence 'composite', (c) Forms cannot 'perish', whereas numbers are conceived as able to do so (104cl—3, 106al). (d) Numbers, unlike Forms, can have contrasted predicates in different relations, e.g. 'more' and 'less', 'double' and 'half'.

Aristotle ascribes to Plato a doctrine of mathematical entities 'intermediate' between Forms and sensible things. The worth of this evidence is disputed, and the doctrine is, at most, inchoate in the dialogues. But an intermediate status for numbers would suit the present parallel between three and soul, in view of the soul's own intermediate status between the Forms and the sensible world (78c— 84b). And the above objections to viewing numbers as (a sub-class of) Forms warrant asking, for each occurrence of 'three' or 'three­ness' in the coming argument, whether a Form or a number is meant. See W. D. Ross, P.T.I. 65-7, 206-12, A. Wedberg, P.P.M.

122-35, J. M. Rist, Phronesis 1964,33-7,1. M. Crombie, E.P.D. ii. 440ff.

For the terms 'three' and 'threeness' see on 101b9—c9. Nothing can be inferred from Plato's terminology. Note, especially, the different locutions used for members of the odd and even number series at 104a4—b4. The casual shifting from 'threeness' and 'fiveness' to 'two' and 'four' in parallel contexts suggests that no systematic distinction is intended between the two types of terms. See also on 104d5-e6,104e7-105b4.

Hackforth (151, n.2, cf.156) argues from the use of'threeness' at 104c5 that'three' at 104cl must mean the immanent Form of Three. But it does not follow from the use of 'twoness' and 'threeness' at 104c5, as Hackforth supposes (151, n.l), that the whole paragraph 104b6—c3 is concerned with Forms. For if'twoness' and 'threeness' at 104c5 mean the numbers two and three, it could well be the number three of which it is said (cl—3) that it will 'perish' rather than become even. Moreover, the items exemplified by three are said (b9—10) not to admit 'whatever Form may be opposite to the one that's in them'. To speak of a Form's being 'in them' is to suggest that 'they' are not Forms themselves.

Whatever may be meant by 'three' or 'threeness', it is difficult to understand the 'military metaphors' in relation to this example. What is meant by the Form Even's 'attacking' three, and how should three's 'getting out of the way' and 'perishing' be understood? As before, it is preferable to take these as genuine alternatives if possible. See on 103cl0—e5 (p.198). Perhaps three's 'getting out of the way' could be understood in a manner analogous to the withdrawal of Simmias' Largeness in face of a comparison with Phaedo instead of Socrates. Three might be thought of as withdrawing from a set, when that set is viewed in terms of a different unit-concept: three 'with­draws' from three musicians when they are viewed as one ensemble; by contrast, it 'perishes' when a fourth member is added to their number. Cf. Parmenides 129c4—d6.

At 104c5 Socrates says: 'Moreover, twoness isn't opposite to threeness.' Hackforth (151, n.3) thinks that this remark has no relevance by itself, and 'restores logic' by supplying in his translation a counterpart statement about two to the one just made about three, i.e. that two will perish rather than become odd. But this is unnecessary. The point of insisting that three is not an opposite is to show that the refusal to admit opposites may be a feature of things that are not opposites themselves. For this is to be the position with regard to soul: it will be held to exclude an opposite, death, even though it is not an opposite itself. Cf. O'Brien, C.Q, 1967,213-15.

104c7—d4. There are major uncertainties of translation in this passage, which affect the interpretation of the whole argument.

The present version of 104c7—9 leaves it open whether the 'other things* mentioned in the second half of tire sentence are Forms. Hackforth, however, translates: 'Hence it is not only two opposite forms that won't endure an onset by one on the other; there are others also that won't endure the onset of opposites.' This makes Socrates claim that not only do Forms of opposites exclude each other, but that certain other Forms likewise exclude one member of such a pair. Accordingly, Hackforth renders the question at 104cll—12: 'Then would you like us, if we can, to specify what sort of forms these are?' However, Forms are not explicitly mentioned in the text, and the question whether the definienda are Forms is prejudged in such a translation. The present version, 'would you like us to define what kinds these are?', leaves the matter open.

At 104dl—3 the translation makes Socrates refer to things that are 'compelled by whatever occupies them' to have not only the occupier's Form, but also that of some opposite. For the grammar and text, see notes 70 and 71. The version adopted follows Tredennick, G. M. A. Grube, C.P. 1931, 197-9, and J. Schiller, Phronesis 1967, 54—5. This is consistent with the trans­lation of 104c7—9 above. For the 'things' referred to here, like those at 104c8, will not necessarily be Forms, but may be other items, which, on being occupied by a Form, such as Threeness, are thereby compelled to have not only the character of that Form, but also that of some opposite Form, such as Oddness. However, the Greek may be equally well, or better, translated either (i) 'those that compel whatever they occupy to have not only their own (viz. the occupiers') character but the character of some opposite as well', or (ii) 'those which compel the object which they come to occupy to have not only its own (viz. the occupied object's) character, but also the character of a certain opposite'. For (i) see O'Brien, C.Q. 1967, 215—16. For (ii) see Hackforth (similarly Bluck). The import­ant difference between both these versions and the translation adopted is that they make Socrates specify as the definienda a class of occupying Forms. For in these versions, the definienda will be identified with the occupiers, and these in turn will be illustrated at 104d5—7 by the Form of Three; whereas on the translation adopted, the definienda will be identified with the things occupied by such Forms, and are therefore unlikely to be Forms themselves.

The present version of 104dl—3 is better suited to the view of the argument preferred in these notes. It enables the disputed items to be taken not as Forms but as physical stuffs or numbers. It is, admittedly, subject to some linguistic difficulty, although in view of the uncertainty of the text at 104d2 this is not decisive (see note 70). In addition, however, one substantive objection to taking 104dl—3 as required by this translation must be acknowledged. It makes Socrates refer to a class of items such that whatever occupies them, they are compelled to have not only the Form of that thing, but also the Form of some opposite. And it may be doubted whether there exists a class of items, such that any Form occupying them imports the Form of an opposite along with it. It is true that in the case of numbers more than one Form may have this effect. It will be mentioned later (105a6—bl) that not'only 'ten' but also 'the double' excludes the Odd. The number ten may thus be thought of as occupied by at least two Forms, those of Ten and Double, that compel it to be even. But it is not clear that all Forms have such an effect upon it.

Possibly, therefore, the phrase 'whatever occupies them' refers not to a plurality of Forms occupying one and the same item, such as ten in the above example, but rather to a plurality of Forms severally occupying the different items that Socrates is seeking to define. These include the whole series of natural numbers (104a7— b4). So Plato, with numbers uppermost in mind, could perhaps have written 'whatever occupies them' with reference to the whole series of Forms for numbers, each Form being thought of as occupying a different number, and making it either even or odd.

It may be useful here to set out alternative versions of the whole argument warranted by the translations just considered. Version A is based on the translation adopted. Version B is supported by rendering the definienda at 104c7—13 as Forms, and translating 104dl—3 with Hackforth or O'Brien. Divergent interpretations of certain passages discussed in later notes will be labelled A or B with reference to these versions. For the term W-instance', used in B3 and B8, see on 102d5— 103a3 (p.196). For an assessment of the key stages of the argument, on each of these versions, see on 105b5—c8 (p.213) and 105c9-dl2 (p.214-15).

Version A

Al. There are items which, al­though not themselves opposites, will not admit one member of a pair of opposites (104c7—9).

A2. These items are such that, if they are occupied by a non-

Version B

Bl. There are certain non-opp­osite Forms, such that what­ever possesses them will not ad­mit one member of a pair of opposite Forms (104c7—9).

B2. These non-opposite Forms are such that whatever they

Version A

opposite Form A, they must poss­ess an opposite Form F(104dl — 8).

A3. Such items will never admit the opposite of the Form F, i.e. the Form G, hence are un-G (104d9-e6).

A4. Such items may be defined as things that bring the Form F to whatever they enter (105a3- 4).

[A3 + A4]:

A5. Items occupied by Form A that bring the Form F to what­ever they enter will never admit the Form G (105a4-5).

A6. If a thing (jc) is such that, whatever y it is in, y will be F, x may be said to 'bring' F to y (supplied).

A7. Soul is such that, whatever body it is in, that body will be living (105c9-d2).

[A6 + A7]:

A8. Soul brings Life to whatever body it occupies (105d3—5).

A9. The opposite of Life is Death (105d6—9).

[A5 +A8]:

A10. Soul will never admit the opposite of what it brings (105- dlO-12).

Version B

occupy must have not only them, but an opposite Form as well (104dl—8).

B3. When a non-opposite Form A brings an opposite Form F to whatever the Form A occupies, then an iV-instance of the Form A will never admit F's opposite, the Form G (104d9-105a5).

B4. If a non-opposite Form A is such that, whatever body (x) it is present in, x will be F, then Form A brings F to x (supplied).

B5. The Form of Soul is a non- opposite Form, such that what­ever body it is present in, that body will be living (105c9— d2).

[B4 + B5]:

B6. The Form of Soul brings Life to whatever body it occu­pies (105d3—5).

B7. The opposite of Life is Death (105d6-9).

[B3 + B6]:

B8. An TV-instance of the Form of Soul, i.e. an individual soul, will never admit the Form opp­osite to that which the Form of Soul brings (105dl0-12).

Version A

A11. What will not admit Death is im-mortal (105dl3—e3).

[A8 + A9 + A10]:

A12. Soul will not admit Death

(105e4—5).

[All + 12]:

A13. Soul is im-mortal (105e6— 7).

A14. What is im-mortal is imper­ishable (106c9-d9).

[A13 + A14]:

A15. Soul is imperishable (106- el—107al).

Version B

B9. What will not admit Death is im-mortal (105dl3—e3).

[B6 + B7 + B8]:

BIO. An individual soul will never admit Death (105e4—5).

[B9 + B10]:

Bll. An individual soul is im­mortal (105e6—7).

B12. What is im-mortal is im­perishable (106c9-d9).

[Bll +B12]:

B13. An individual soul is im­perishable (106el-107al).

104d5—e6. Socrates illustrates the suggestion made at 104dl—3 with the example of three. This passage contains the one and only unambiguous mention of a Form, other than Forms of opposites, in the entire argument. Those who believe that the items under dis­cussion are Forms will naturally regard the Form of Three (d5—6) as an instance of the class that Socrates had proposed for definition at 104cll—12. The passage is important, in view of the deliberate parallel between the cases of three and soul, and the similarities of language with 105d3-5. But it does not settle the question whether the definienda are Forms. For, on the translation adopted at 104dl—3, they will be exemplified here not by the Form of Three, but rather by whatever it is supposed to occupy (see previous note). If so, the ,passage will be identifying the definienda not with Forms that import opposites, but with the items occupied by such Forms. Its argument will then be that those items will not admit the Form opposite to the one imported into them. This will not constitute a 'definition' of the items in question. Indeed, it is not until 105al—2 that Socrates says 'see if you define them thus'.

Here the problem of interpreting Plato's locutions for numbers again becomes acute. D. O'Brien has argued (C.Q. 1967, 216—19) that 'in the elaboration of the numerical example (i.e. from 104b6) there is a consistent distinction between form and particularisation. The form is described as "the Form of Three" or more simply as "threeness". The particularisation is described as "three".' But this alleged distinction is very hard to sustain. Both at 104c5 and 104e5, 'threeness' follows immediately upon occurrences of 'three', where the latter stands, in O'Brien's own view, for the number three. The point of switching from number to Form in these places is difficult to see. Moreover, as O'Brien recognizes (op.cit. 218), 'threeness' is used in parallel with 'two' at 104a4—b2, where 'threeness' and 'five- ness' exemplify the odd series of numbers, and 'two' and 'four' the even. O'Brien suggests that it is natural for Plato's language 'to become firmer with the elaboration of his example'. But Plato could quite well have registered this important distinction, had he wished to, from as early as 101c5—6, where the '-ness' terms were used for the Forms of One and Two. Again, if the distinction is supposed to be firmly established from 104b6 onwards, why does Socrates back­slide at 105a6—7? See next note (p.208).

Hackforth (151, n.2, 152, n.l, 156) holds that 'three' and 'three­ness' are used interchangeably to stand for the immanent Form Threeness, and that the meaning of 'three' at 104el—3 can be fixed from the use of 'threeness' at 104e5. In drawing no distinction between 'three' and 'threeness' he seems correct. But it does not follow that both are used to mean 'immanent Form'. It seems equally possible to hold that (i) neither 'three' nor 'threeness' means a Form of any kind, but that (ii) both alike refer to the number three; and that therefore (iii) the only reference to the Form of Three in the whole passage occurs in so many words at 104d5—6.

But even if 'three' and 'threeness' are taken as just suggested, the meaning of 104d5-7 and the point of the succeeding argument are still uncertain. For what is it that is supposed to be occupied by the Form of Three and compelled to be not only three but also odd? Is it the number three, or is it sets of three things? O'Brien (op.cit. 212, 217) argues that only the number three could be said to be odd 'by nature' (104a3, a7); a group of oxen, which might change in number, could not. But the claim that whatever the Form of Three occupies must ipso facto be odd seems as plausible for sets as for numbers. Cf. Hippias I, 302al—7. O'Brien also appeals to Socrates' later reference to a 'number's' being made to be odd (105c4). But this is not conclusive, since 'number' may mean a numbered set (cf. Phaedrus 247a2). It is therefore unclear whether Plato is thinking of the Form of Three as (A) occupying the number three or as (B) occupying a set of three things.

There are, accordingly, two ways of taking the argument of this section. (A) 'Whatever the form of three occupies' refers to the number three, and it is to this that 'a thing of that kind' refers back at 104d9. On this view, the argument runs as follows: (i) whatever the Form of Three occupies will be odd; (ii) the Form of Three occupies the number three; hence (iii) the number three will be odd, and hence (iv) uneven. This interpretation would suit Version A of the argument given in the previous note. Cf, J. Schiller, Phronesis 1967, 57-8.

Alternatively, (B) 'whatever the form of three occupies' may be taken as sets of three things, and the argument understood as follows: (i) whatever sets the Form of Three occupies must be odd (d5—8); hence (ii) the number three will have no part in the Form opposite to the Odd, i.e. the Form Even (d9—e4); hence, (iii) the number three will be uneven (e5—6). Here, as in Version A, 'a thing of that kind' (104d9) refers back to the number three. But the number is viewed as an iV-instance of the Form (see on 102d5—103a3, p. 196); and the argument runs from the effect exerted by the Form of Three upon its ^-instances, triadic sets, to the property of the number three: the number, or iV-instance, will have essentially that property which the Form compels the sets to have while it occupies them; it will therefore not admit that property's opposite.

The general principle implied on this interpretation would be that when a non-opposite Form A compels its T-instances to possess an opposite Form F, then aniV-instance of the Form A will never admit the Form G (see B3 in the summary of the previous note). This principle would be applied to the soul by treating it as an TV-instance of the Form of Soul (i.e. as 'the soul in us'), and the bodies it animates as ^-instances of that Form (i.e. as 'things besouled'). See B6 and B8.

On either of the above versions, the parallel between the number three and the soul is dubious. For it is hard to understand either the Form's occupancy of the number three (Version A), or the treatment of the number three as an TV-instance of the Form (Version B), if there is only one such number. The referent of 'the number three' is naturally taken, in English at least, to be unique, whereas that of 'the soul' need not be: 'the soul' may mean 'our souls' (just as 'the appendix' may mean 'our appendices'), whereas 'the number three' can hardly stand for a plurality. If Plato is taken to have believed in a plurality of threes, as in the doctrine of intermediates referred to on p.200, it would be possible to think of'a three' rather than of 'the number three'. But such expressions as 'a three' or 'threes' are more naturally used in English of sets than of pure numbers. The required parallel between 'three' and 'soul' is therefore peculiarly difficult to express.

104e7—105b4. This difficult passage divides into two parts: (1) At 104e7— 105a5 Socrates proposes a definition of the class of entities he had suggested defining at 104cll-12. (2) At 105a5-b3 he gives some further numerical examples.

(1) For the complex grammar of the 'definition' (a2—5) see note 74. It does not, unfortunately, read like a definition at all, but simply enunciates the general principle that whatever A items bring F to something cannot themselves admit G. This is a cardinal principle of the argument, which will be applied directly to the case of soul at 105dl0—11. See on 100e5-101b8 and next note. Its meaning, however, turns upon whether the designated A items are Forms, and this still remains unclear.

At 104e8—105al they are exemplified by 'threeness', 'twoness' and 'the fire'. The translation 'the fire' keeps the Greek definite article, but should not be read as a definite description, referring to an individual fire. It means either 'fire' in a generic sense, i.e. stuff called 'fire', or the Form of Fire. D. O'Brien (C.Q. 1967,220) argues from the numerical locutions 'threeness' and 'twoness' that Plato is probably 'now thinking of fire to some extent as form'. But the numerical language is quite inconclusive (see on 101b9-c9 and previous note). In the further examples (a6—7) of what is, presum­ably, the same point, Socrates switches to the ordinary terms for the cardinals 'five' and 'ten'. This confirms the view of J. Schiller (Phronesis 1967, 57) that Plato shows a 'studied indifference to the locutions by which he refers to numbers'. It cannot, therefore, be argued either, on the strength of 104e8—105al, that Forms are meant, or, on the strength of 105a6—7, that they are not. Hackforth's translation at 105a3—4 begs the question by importing Forms where there are none in the Greek.

At 104el0 and 105a3—4 Socrates speaks for the first time of the items in question 'bringing' an opposite to something. The same word is used later (105d3-4, d 10-11) of soul's 'bringing' life to body, and may be an extension of the military metaphor, referring to the 'bringing up' of reinforcements. However, the meaning of 'bring', and consequently the identity of the 'bringer' and the 'recipient' of an opposite Form, are disputed.

(A) 'Bringing' may be understood as 'causally imparting', and the 'bringer' may be taken not as the Form A, but as any x that participates in the Form A, and is therefore F, and by its presence imparts F to another individual, y. On this view, the principle being formulated here will be:

(xjO {(Ax.x causesy to be F) "D (~Gx)}. If this formula expresses Plato's meaning, its application will be as follows: 'A' will be replaced by 'participates in the Form Soul', and 'G' by 'participates in the Form Death', x's causing y to be F will represent an individual soul's 'bringing' life to any body that it enters, i.e. 'causing' it to be alive, and it will be from this that soul's refusal to admit death will be inferred. This fits Version A given at 104c7— d4 (p.204).

Alternatively, (B) 'bringing' may be taken to stand for the same relation as was illustrated at 104d by the Forms Three and Odd, i.e. a non-symmetrical relation between two Forms, A and F, such that whatever participates in Form A must also participate in Form F. Cf. G. Vlastos, P.R. 1969, 317. On this account, the 'bringer' will be a Form, such as Three, and the 'recipient' will be an individual that participates in that Form. Thus, where A is the 'bringing' Form, F the Form brought, and G the Form opposite to F, the principle being enunciated here would, on this view, be simply: (x) (Ax D ~Gx).

Note, however, that this formula would not fit Version B of the argument, and would not yield the premiss required for proving the soul's immortality. For Socrates will not argue that it refuses to admit Death merely on the ground that it participates in the Form Soul, i.e. that it is soul. He argues that it excludes death because it brings life to the body. If, therefore, the argument is to be construed in terms of entailment relations between Forms, the more complex pattern of argument given under (B) in the previous note (p.207) will be needed.

(2) The further numerical examples are well explained by D. O'Brien, op.cit. 221-3. See also F. M. Cornford, C.Q. 1909, 189—91. The words 'This, of course, is itself also the opposite of something else; nevertheless, it won't admit the form of the odd' (a8—bl) are difficult. They seem meant as an aside about 'the double' as such, and not about ten qua double. Hackforth (153, n.l) translates and interprets the second half of the sentence as if its subject were 'ten'. But it seems better to take the subject as 'the double' throughout. The point will then be that 'double', unlike the numbers two, three, five, and ten, that have been instanced as excluding odd or even, is itself an opposite of something else, namely 'half' (cf. Republic 438cl—2, 479b3); nevertheless—i.e. des­pite its infringing the norm by being an opposite—it still excludes the odd, since no number that is double can, in fact, be odd.

Two series of fractions are instanced (bl—3) as excluding the Form of Wholeness: 3, §, f, etc., and 3, %, etc. These examples are of no importance for the main argument.

105b5—c8. Socrates now puts forward a new kind of answer which he illustrates with a series of examples, 'fire', 'fever', and 'oneness' (or 'a unit'). The main problems here are: (1) What exactly is the grammar and sense of the lines indicating fire and fever as the reasons for a body's being hot or ailing (b8—c4)? (2) What is the meaning of the numerical example? (3) Flow is the new type of answer related to the old 'safe' answer in terms of Forms? (4) Why

does Socrates see in it 'a different kind of safeness'?

For the grammar and text at 105b8~c4 see note 75. Hackforth translates 105b8—9: 'what must be present in a thing's body to make it hot?' (with parallel renderings of the questions at 105c3—5). But this wrongly suggests that fire and fever are necessary conditions for heat and illness respectively; whereas fever, at least, is clearly not a necessary condition for illness but only a sufficient one (cf. Alcibiades II, 139e—140a). Fire, too, is probably not meant as a necessary condition for heat, since it is parallel in the argument to snow, and the latter could not be held a necessary condition for cold (note also 63d7—8). Hackforth (161) regards it as a weakness of the answers 'fire' and 'fever' that they give conditions that are merely sufficient and not also necessary. Of course, if the relevant phrases are translated as if they specified necessary conditions, they will appear to be making a false claim. But they clearly should not be so translated. It is only in the case of soul (105c9—d2) that the new answer gives a condition that is indisputably necessary as well as sufficient. In general, then, the new answers differ from the old ones in being merely sufficient, whereas the old ones were both sufficient and necessary.

The word translated 'oneness' (morns) at 105c6 may stand for the Form of One, as it clearly does at 101c6. But it may also be rendered 'a unit'. Hackforth translates it thus, and explains it (158, n.2) as 'the one left over in the middle when an odd number is divided into two equal parts'. This explanation conforms to two definitions of 'odd' criticized by Aristotle in the Topics: 'that which is greater by one (morns) than an even number' (142b8), and 'a number with a middle' (149a30—31). See also Euclid, Elements vii, Def. 7, ed. T. L. Heath, ii. 281, and H. F. Cherniss, ACiM. 25, n.19. Hackforth's interpretation gives a clear sense in which the 'odd' or unpaired unit makes a number odd, and is well suited to the root meaning of the Greek word for 'odd'. Cf. also O.E.D., s.v. 'odd'.

Thus understood, 'a unit' would be both a necessary and sufficient condition for three, five, etc. being odd. It would then be out of line with the cases of fire and fever, but in line with that of soul. Hackforth's interpretation, however, seems inconsistent with his further account (162), in which 'Unit' is treated as a Form, importing Oddness. For it is hard to see how 'a unit', as Hackforth explains it, could be thought of as a Form. If a Form is meant, it must be the Form of One, thought of as bringing oneness to any particular 'one' that it occupies, and thereby making it odd. In that case the translation 'oneness' will be required, and the example would be in line with the cases of fire and fever, giving a sufficient, but not a necessary condition for oddness. Cf. O'Brien, C.Q. 1967,

224-5.

Would 'a unit' itself be odd? W. D. Ross (ed. Aristotle, Physics, 604) says that according to the normal view of Greek mathe­maticians, two was the first number. On this view, 'a unit' would not, strictly, be a number, but that of which numbers consist, or in terms of which 'number' is defined. See also T. L. Heath, H.G.M. i. 69-71, and M. E. Hager, C.R. 1962, 1-2. The 'unit' (morns) was sometimes called 'even-odd', being thought of as a parent of all numbers. This, if relevant here, would create some difficulty for the principle that a 'reason' must exclude the opposite of that for which it is a reason. But, a different, perhaps more popular, conception of 'one' (to hen) is found at Hippias I, 302a3—5, where it is explicitly said to be odd. It must be so regarded here on any interpretation of the argument. It remains unclear, however, whether what is made odd is regarded as a pure number or as a single-membered set. This turns on the meaning of 'number' at 105c4. See on 104d5— e6 (p.206),

(3) As noted above, the present series of answers give sufficient, but not necessary conditions. They do not, therefore, satisfy the principle, implicit in the earlier discussion, that no opposite, F, can count as the reason for something, if its opposite, G, can give rise to the same thing (see on 100e5—101b8, p.186). Thus, the answer 'fever' is open to the very objection that Socrates raised against 'addition' at 97a7—b3: illness could be due to the opposite of fever, hypothermia. Evidently, therefore, the present 'reasons' are not meant as constitutive of what they explain. They do not aim at answering the conceptual question 'what (logically) makes things F?'

What, then, is their purpose? There are two different ways of interpreting them, according as they are or are not construed as Forms.

(A) If they are not taken as Forms, they will specify reasons that 'make' whatever they are in to be F, in a causal and not merely a logical sense of 'make'. On this account, the new answer does not supersede the old 'safe' one in terms of Forms, but supplements it, by showing how a particular object or number comes to be occupied by the Form in question. Thus, fire makes bodies participate in Hotness. Similarly, fever, regarded as a cause rather than as a symptom of illness, makes them participate in Illness. This account fits the numerical example less well, since the notion of causal 'making' is strictly inapplicable here. But an unpaired unit may be thought of, analogously, as imparting a Form, Oddness, to a number or set that participates in that Form. Soul will be thought of, similarly, as imparting the Form Life to the body that it occupies (105c9—dl2). It 'makes' the body to be alive, in no mere logical

sense of 'makes', but in the sense that it gives it life, or quickens it.

Alternatively, (B) The reasons here specified are Forms of Fire, Fever, and Oneness. Thus, with variations, Hackforth, 162, G. Vlastos, P.R. 1969 , 317-20, D. O'Brien, C.Q. 1967, 223-8, E. L. Burge, Phronesis 1971, 11—12. Understood thus, the new answers specify, not 'causes' of a thing's being F, but logically sufficient conditions from which its F-ness may be inferred.

Once again, no conclusions can be drawn from Plato's term­inology. There is no explicit reference to Forms in the passage, unless it be held that at 105c6 'oneness' must be meant, in conformity with 101c6. But no decisive inference can be based upon the locutions for numbers (see on 104d5-e6, p.205—6).

If the 'reasons' given here are taken as Forms, the 'bodies' in which they are said to be present cannot, unless Plato's examples are hopelessly disparate, simply be //-instances of the Form in question, e.g. a particular fire, thought of as occupied by the Form of Fire. For such an interpretation would not fit the fever example. Socrates could not refer to the presence of the Form Fever in a particular fever as its presence in a 'body' (c3), for a particular fever, unlike a particular fire, is not itself a body. If, therefore, Forms are meant here, the 'bodies' they occupy cannot be fires, but must be things on fire, e.g. sticks. Nor can they be fevers; they must, rather, be things feverish, e.g. human bodies. Similarly, in the case of soul: Socrates could not be referring at 105c9-ll to the Form of Soul's presence in a particular soul. For the soul is not a body. In view of this, the Form interpretation of these lines can be made to fit the case of Soul and Life at 105c9-dl2 only with great difficulty. See next note.

(4) Some of the new answers of this passage, such as fire and fever, may seem no less 'mechanistic' than the ones earlier rejected (see C. C. W. Taylor, Mind 1969, 52-3). Indeed, by calling them 'subtler' (c2), Socrates links them with the answers that were unacceptable before (cf. 'subtleties' 101 c8, and 'those other wise reasons' lOOclO). Why, then, are such 'subtleties' now admitted as giving 'a different kind of safeness'? How could they be thought 'safe' from the dangers that had threatened such answers before?

As noted in (1) above, they do not enjoy the complete safeness of the old Form 'reasons', in that they are not necessary conditions for what they explain. But they do satisfy Socrates' other require­ments for 'reasons'. In particular, and of paramount importance for the argument, they do not admit the opposite of the properties that they impart—e.g. fire cannot itself be cold. This what makes them 'safe' as well as 'subtle'. It is this principle that underlies the proof that soul is immortal: since it is the 'reason' for life, soul cannot .itself admit death, and therefore cannot be dead (see also on 100e5—

101b8, p.186—7).

What can be said of this principle? It has some plausibility in such cases as fire and snow, which may be thought of as trans­mitting their own heat and cold to other bodies. Clearly, they could not do this, if those properties were 'neutralized' by the presence of their opposite. But the principle is less plausible in the other examples. Fever does not transmit illness from itself to the body in which it is present. It is an illness, but is not itself ill. Nor could the principle be held to apply to reasons more generally, (i) It would have no application to properties that are not members of a pair of opposites. (ii) Even where the properties to be explained are members of such a pair, it is untrue that a reason for one of them must exclude its opposite. A germ that is a 'reason' for illness in a body may itself be healthy. A saccharine pill that is a 'reason' for sweet­ness in coffee may itself be bitter. But fundamentally, (iii) in any given case it may be doubtful whether there is any 'reason' to be found that satisfies the principle under discussion. There seems nothing analogous, in this respect, to fire and snow that could be called a 'reason' for a thing's being large or small. In the case of life, 'soul' is, of course, taken to be a reason that meets the con­ditions stipulated. But this simply prejudges the question whether, in fact, life admits of explanation in terms of the kind of 'reason' that Socrates has displayed. Unless it does, his final proof of immortality cannot get off the ground. See next note.

105c9—dl2. Soul is here specified as the reason for body's being alive (c9—11). Cebes does not balk at this. Presumably, it fits his own conception of soul as 'making' body. See on 87d3—e5.

Burnet's text, which gives 'soul' without the article throughout 105cll—e6, has been followed, although the MSS. vary on this point at 105d3, dlO, e4, and e6. 'Soul' without the article may mean 'soul-stuff rather than individual soul. See on 64e4—65a3 (p.89). Some commentators have wished to understand it, rather, as the Form of Soul, as at B6 in Version B given in the note on 104c7-d4 (p.204). The difficulties to which this leads may be illustrated from G. Vlastos's account of the context (P.R. 1969, 317-20).

Vlastos interprets the inference pattern established by the pre­ceding examples as: 'x is F because, being A, it must participate in the Form A; and since the Form A entails the Form F, x must also participate in the Form F, and hence x must be F.' (His variables have been adapted to the conventions followed in these notes, where F and G are used for the Forms of opposites, and A for the Form of a non-opposite). Vlastos does not discuss the application of this formula to the argument for immortality. But it is clear that an argument of the proposed pattern could not prove the immortality of the soul in the way required by the present text. For applying the formula, we should have either:

(1) 'A soul is alive because, being a soul, it must participate in the Form Soul, and since the Form Soul entails the Form Life, a soul must also participate in the Form Life, and hence a soul must be alive.'

or (2) 'A body is alive because, being besouled [or being a soul] it must participate in the Form Soul, and since the Form Soul entails the Form Life, the body must also participate in the Form Life, and so must be alive.'

But neither (1) nor (2) is satisfactory. (1) will not fit the text, for it ignores the soul's relationship to the body, which cannot be eliminated, even if the 'subtle' answers of 105b—d are taken as Forms (see previous note). And (2) clearly does not yield the conclusion for which Socrates wishes to argue. He needs a conclusion not about body but about soul. Moreover, he needs a con­clusion not just about the Form of Soul, which is 'immortal' like any other Form, but about a particular soul. Yet if the reference at 105c9—d5 is to the Form of Soul, it is hard to see where or how the transition to particular soul, or soul-stuff, is supposed to occur. In Version B a transition has been effected at 105dl0—11, by taking those lines to mean that an individual soul will never admit the opposite of what is imparted by the Form of Soul to the body (step B8). But this reading is factitious, and not warranted by anything in the text.

Note also that the words 'Then soul, whatever it occupies, always comes to that thing bringing life' (d3—4) seem intended as an inference from 105cl 1—d2. So if the earlier lines refer to the presence of particular soul, or soul-stuff, in the body, it is natural to take 105d3—4 in the same way.

It is therefore preferable to take the whole of the present passage as referring to a particular soul, or soul-stuff, whose presence in a body quickens it, as in Version A. Whether or not Platonic ontology recognizes a Form of Soul, parallel to the Form of Three at 104d5— 6, and despite the parallelism of language with that passage, a more coherent argument emerges if such a Form is not read into the present text.

On either of the versions distinguished, the main principle of the argument is vulnerable. On Version A, the critical point is the claim that items 'bringing' an opposite F to whatever they enter will not admit G. Since this principle (A5 in the summary at 104c7-d4) depends upon the 'definition' at A4, the argument is flawed by defining the entities in question as those that 'bring' the Form F to whatever they enter. It may, of course, be simply stipulated that no x shall count as a 'reason' for _y's being F unless x is itself un-G. But if this artificial restriction is to be imposed upon the concept of a 'reason', it has to be asked whether there might, for some values of F, simply be no 'reason' for things' being F that meets so stringent a condition. See previous note (p.213).

On Version B, the core of the argument will be (B6—B9) that since any body occupied by the Form Soul must be alive, no individual soul can ever admit death. The mainspring of this inference will be the principle (B3) that TV-instances of a Form A will not admit the opposite of a property F imparted by that Form to its ^-instances. But this too seems questionable. For TV-instances may, perhaps, be qualified by the opposites of the properties their Forms impart. An individual hemlock plant, an TV-instance of the Form Hemlock, may be living, even though jT-instances occupied by the Form of Hemlock (i.e. bodies sufficiently dosed with hemlock) must be dead. This example may seem contrived. But the conception of the Form of Soul's occupancy of a body (B6) is itself both artificial and opaque.

105dl3—e9. It is now argued that since soul will never admit the opposite of the Form that it brings, i.e. the Form of Life, it will not admit death, and must therefore be 'im-mortal'.

At 105dl6—el the words rendered by their derivatives, 'musical' and 'un-musical', have the broader sense of 'cultured' and 'uncul­tured' (see on 60c8—61cl). 'Un-musical' and 'un-just' are used, like 'un-even' (dl5), for verbal symmetry with 'im-mortal'. Similarly, at 106a3—10 'un-hot' and 'un-coolable' translate words coined by Plato to parallel 'im-mortal'. It is important to take these predicates as meaning not merely 'not being G' but 'never being G' or 'incapable of being G\ This is the point of inferring 'is un-G' from 'does not admit the Form G'.

Failure to take note of this has sometimes led to charges of equivocation upon the word for 'immortal'. Thus, D. Keyt has argued (Phronesis 1963, 170—1) that in the present passage it means merely 'not dead', or 'alive', whereas by 106e5—6 it has come to mean 'never dying'. But this disregards the force of 'doesn't admit death', from which 'im-mortal' is derived at 105e2—3. To say that a thing does not 'admit' G means that it will never, while remaining itself, participate in that Form. Cf.l02e, 103d, 104b-c, 105a, and note the emphatic construction translated 'will absolutely never admit', at 105dl0—11. It follows that the predicate 'im-mortal' means not merely 'is not dead' but 'will never be dead'. There is, therefore, no shift in the meaning of 'immortal'. It means con­sistently 'never dying' or 'incapable of dying'. See also on 64c2—9 (p.87), 72e7—73a3 and 95b5-e6.

Keyt interprets 'soul doesn't admit death' at 105e4 as 'nothing can at one time possess a soul and be dead', and then objects (op.cit. 171) that in order to get his desired conclusion Plato must sub­sequently shift the predicate 'immortal' from that which has soul to soul itself. But this reading of 'soul doesn't admit death', which stems from construing 'soul' at 105d3—5 as an'immanent Form', is unnatural. If 'soul' is not taken as immanent Form, but as particular soul, or soul-stuff, then the predicate 'immortal' can readily attach to soul itself at 105e6, and no shift of subject need be supposed.

The charge of equivocation upon 'immortal' is related to another, often repeated criticism. The argument proves, it has been objected, only that soul is not dead 'whenever it exists' or 'while it is still soul' or 'so long as it is': it does not prove that soul is not at any time dead, and so not-dead or 'immortal' in the required sense. D. O'Brien (C.Q. 1967, 229-31) suggests that Plato may think he is entitled to omit the qualification 'whenever it exists', in virtue of his contention that soul is 'always', i.e. necessarily or essentially, alive. But it seems more likely that he was aware of the need for proof of the existence of a subject of the required type, and that this is precisely what gives rise to the attempted proof of soul's 'imperishability' at 106a—e. See next note (p.217). At Euthydemus 296a—d it is noticed that 'always' is logically treacherous, and that 'whenever' clauses may not safely be dropped after its occurrence.

The concept of immortality is not, however, without difficulty. For what is meant by the 'death' which soul will not admit? It cannot here mean 'separation from the body', for soul will admit 'death' in this sense (see on 64c2—9, p.87). Could it mean 'perishing'? This is also difficult, for two reasons, (i) Death has just been said to be the opposite of Life (d6—9), and must therefore be thought of here as a state rather than as an event, whereas 'perishing' is an event, (ii) If 'death' means 'perishing', what is to be made of the claim that soul is immortal and imperishable, and of the coming argument for the latter predicate? It is, in fact, hard to understand 'death', and 'immortal', in such a way that the final page of the proof is neither otiose nor question-begging. Related difficulties about the meaning of 'death' will arise at 106bl—4 and 106e5—7.

105el0—107al. This section contains the final phase of the argument. Its purpose is to distinguish the case of soul from those of snow, fire, and three, to which it has so far been assimilated. In those cases, the possibility of 'perishing', in face of ail attack by the excluded opposite, remained open. But in the case of soul, this possibility is to be ruled out; soul is not only immortal, it is also imperishable.

The point of contrasting soul with the other cases is to fill a gap in the proof of immortality expressed by Hackforth as follows (163): 'what has been shown is that the predicate "deathless" is contained in the meaning of the subject, soul; whenever, therefore, this subject exists, it has this predicate: but to show that the subject always does exist is quite another matter.' See the criticism of Strato of Lampsacus, given by Hackforth, 196 (g) and (m), and previous note. The present phase of the proof is meant to forestall such criticism. It tries to move beyond the tautology 'as long as a soul exists, it is not dead' to the claim that soul does, indeed, always exist. Like the Ontological Argument, it seeks to conjure an existential proposition out of a conceptual analysis. This is the reason for insisting upon 'imperishable' as an additional predicate (cf.88b6,95cl, and see on 88al-b8,95b5-e6).

In outline the argument runs as follows:

If the immortal is imperishable, soul will be imperishable

(c9—dl).

The immortal is imperishable (d2—9).

So

Soul is imperishable (el—107al).

There are uncertainties of interpretation at each of these stages.

1.(1) Does 'the immortal' mean (i) 'that which is immortal', or (ii) 'the property of being immortal'? For (i) see D. O'Brien, C.Q. 1967, 207. For (ii) see D. S. Scarrow,P.R. 1961, 245-52. With© the argument will be that since whatever is immortal is imperishable, the soul, being immortal, must be imperishable as well. With (ii) the argument will be that since the property of being immortal is imperishable, its bearer must be imperishable as well.

(ii) is a possible reading, if 'the immortal' is construed as parallel with 'the odd'. For 'the odd' is used at 106b5—6 in a manner parallel to 'the hotness in the fire' (b6—7). At 106b7—8 also it means 'oddness' rather than 'that which is odd', for at 106c5, which looks back to 106b7—8, the meaning must be 'oddness and three depart and go away'. But against this reading of 'the immortal': (a) It is far from clear that 'the im-mortal' is being used in a manner parallel to 'the odd'. Formally, it is parallel, rather, to 'the un-even', the 'un-hot' and 'the un-coolable'. These probably mean 'that which is un-G' rather than the property 'un-Gness'. (b) 'The immortal is also imperishable' at 106b2 and 106c 10 (cf.el) seems more likely to mean 'that which is immortal is also imperishable'. For on this view, the point of 'also' will be that the subject is imperishable as well as immortal, whereas on the alternative view 'also' has no clear function at all. (c) At 106e6 the phrase translated 'the immortal part' must mean 'that which is immortal', since it alludes directly to the soul. Scarrow's reading would therefore involve a shift at some point in the meaning of 'the immortal'. And (d) the allusion to God and the Form of Life (d5—9) would remain, on this view, as a rather lame pendant to the argument.

The references to 'death' at 106b3 (cf.e5) raise several problems: (a) 'Death' can hardly be understood here either as (i) 'separation of soul from body' (64c2—9), or as (ii) 'perishing of soul' (91d6—7). Clearly, in the words 'it won't admit death nor will it be dead', 'death' cannot be understood in sense (i). For the soul does 'admit death' in that sense of 'death', when it is separated from the body (see III below). O'Brien (C.Q. 1968, 101, n.2) suggests that (ii), which he takes to be a redefinition of 'death', enables 'nor will it be dead' (b4) to mean 'nor will it cease to exist'. But this is not satisfactory either. For Socrates is here recalling the conclusion of 105e2—7, that soul is immortal. Since this conclusion is a premiss in his present argument for its imperishability, he could not use the words 'nor will it be dead' to mean 'nor will it cease to exist', without assuming exactly what he is trying to prove.

Both here and at 106e5—7 it is difficult to interpret the notion of death's 'attack'. Note the difference between this and the impact of hot and cold upon snow and fire at 106a. It is not the opposites themselves, hot and cold, that are said to attack snow and fire, but something characterized by those opposites. But there is no embodiment of death in something that is 'applied' to the soul, or to the man, as something hot or cold is physically applied to snow or fire (106a4, a9).

Did Plato postulate a Form of Death? Although such a Form seems required by the framework established for the argument, Socrates nowhere speaks of one in so many words. A Form of Death, unfortunately, generates an awkward paradox. If included in the Form-world as characterized earlier (79d2, 80bl, 81a5), it would have to be immortal or 'deathless'. But a deathless Form of Death would seem to infringe the basic principle of the present argument, that no Form will admit its own opposite.

At 106b7—cl Bluck translates: 'what is there to prevent it (sc. the odd) from perishing and in that way ceasing to be odd, and becoming even?' (his italics). But this would involve the paradox of saying that the odd, having perished, acquires the fresh property 'even'. This recalls the earlier problem of finding a subject for 'comes to be G' where G is a numerical predicate. See on 96e6—97b7 and notes 58 and 65. It is easier to take 'even' at 106cl as the subject of 'coming to be', or, as in the translation, to render the phrase impersonally. Cf. O'Brien, op.cit. 95, n.3.

II. It is natural to take 106d2—9 as a proof that 'the immortal is imperishable'. For the meaning of 'the immortal', see 1(1) above.

One way of taking the proof is as follows:

If the immortal, being everlasting, admits perishing, then

there could hardly be anything that does not admit it (d2—4).

But

There are things that are imperishable, viz. God and the Form

of Life, and anything else immortal there may be (d5-7).

So

The immortal is imperishable.

On this view, the function of (2) is to deny the consequent in (1) by affirming the existence of some imperishable things. This would enable the antecedent in (1) to be denied by contraposition. And this denial would be equivalent to (3). But on this reading, the words 'anything else immortal there may be' in (2) would be question- begging. For it would be illegitimate to adduce 'anything else immortal there may be' as cases of immortal things that are imperishable, in order to prove that immortal things are, as such, imperishable.

But it is also possible that (1) and (2) should be taken separately, rather than as premisses in a single argument. On this view, the real argument resides in (1). Cebes is there implying that the immortal must be regarded as imperishable, inasmuch as it is everlasting. He is assuming, without argument, that there are, indeed, imperishable things. Socrates then endorses his point by mentioning items that are, in fact, both immortal and imperishable, viz. God and the Form of Life.

But if (1) is read in this way, the phrase 'being everlasting' seems to beg the question. For whether the immortal is, indeed, 'ever­lasting' is precisely what is at issue. Moreover, the words 'anything else immortal there may be* are still difficult. For what could they refer to? Individual souls could hardly be meant, for Socrates would then be anticipating the conclusion drawn at 107al. On the other hand, if Forms, other than the Form of Life, are meant, why is the Form of Life itself singled out for special mention? Are not all Forms equally imperishable?

For the argument to work at all, it must be supposed that there are at least some things that do not admit perishing. What basis is there for this supposition? It is sometimes suggested that Plato is here relying upon the unstated assumption that 'nothing can come from nothing or disappear into nothing'. See, e.g., Archer-Hind, 119, n.l. The possibility of everything's perishing and nothing's existing is, on this view, being implicitly ruled out. The postulate that there must always exist something is being taken as fundamental, like the earlier, undefended assumption that not everything could finish up dead (72c5— d3).

But such a postulate would seem inadequate for the purposes of the argument. For the supposition (SI) that there must always exist something should be distinguished from the supposition (S2) that there must be something that always exists. The postulate just mentioned asserts only SI. Yet it is S2 that is required. Even if SI were granted, would it afford any support for S2? And would S2 be effectively supported by invoking God and the Form of Life?

Some commentators have supposed that Plato here falls back upon an appeal to religious faith. Thus Hackforth writes (164): 'it is only if we allow that the appeal is to faith that we can avoid a feeling of deep disappointment in this matter, inasmuch as from the standpoint of logic the argument has petered out into futility.' But the bathos of an appeal to religious faith at the climax of a philosophically sophisticated argument would only deepen the disappointment. A lame appeal to 'divine doctrine' is the last thing we should expect in the light of 85d2—4. Moreover, when Cebes says (d8—9) that the imperishability of God and the Form of Life would be admitted 'by all men', he cannot be serious. God's imperishability would not be admitted 'by all men'. A sceptic would object that it is at least as much in need of proof as the immortality of the soul. The 'religious faith' interpretation simply undercuts the central enterprise of the dialogue.

It is, however, possible to understand the allusion to God and the Form of Life differently. For these two entities are precisely the ones required, in terms of the kinds of 'reason' that Socrates has displayed, as 'reasons' for the existence of the universe as a whole. In the case of living things, 'life' and 'existence' are plausibly identified (see on 95e7—96a5, p.171); and the universe as a whole is viewed by Plato as a living thing (see on 72all—d5, p.l 13). The 'safe' reason for a thing's being alive is that it participates in the Form of Life. And the corresponding 'subtle' reason is that some­thing else has brought Life to it. At the level of individual organ­isms, this will be soul. And at the cosmic level, it will be God.

A possible role for God and the Form of Life, within the framework of the foregoing argument, thus suggests itself. They supply, respectively, the subtle and safe reasons needed to explain the existence of a living universe. In this role, God constitutes the cosmic 'reason' for which Socrates has searched (99c6—8). Such a 'reason'answers, in metaphysical terms, his initial question (96b2—3) about the source of life. It fits the allusions to God's designing intelligence in his criticism of Anaxagoras (see on 97b8—98b6), and anticipates the theistic account of the universe in the Timaeus.

If this is correct, the allusion to God is no mere deus ex machina, but an application to the universe as a whole of the pattern of explanation implicit in Socrates* earlier examples. It may even be said to point to a quasi-causal argument for the existence of an eternal being. The pivotal principle of such an argument would be, once again, that a true 'reason' cannot admit the opposite of that for which it is the reason (see on 100e5— 101b8, p.186, and 105b5—c8, p.212). If the living universe demands an explanation of that sort, then there must be something that cannot admit the opposites of life and existence, and can therefore neither die nor perish. Such a reason is God, the true 'reason', perhaps, that Socrates has said he will 'display' (100b8). In this way his 'second voyage' turns out, after all, to lead to the kind of 'reason' he had hoped to discover. And the scholastic arguments for a necessary and eternal being can be seen as 'footnotes' to Plato's text.

III. At 106e5—107al the argument is clinched: 'when death attacks a man, his mortal part, it seems, dies; whereas the immortal part gets out of the way of death, departs and goes away intact and undestroyed.' Here, as at 106bl—4, there is difficulty in interpreting 'death'. It cannot be equated either with (i) 'separation of soul from body' (64c2—9) or with (ii) 'perishing of soul' (91d6—7). Substitution of (ii) for 'death' leads to con­fusion. 'Perishing of soul' could not explain the word 'dies', which is applied at 106e6 to 'his mortal part', i.e. the body; nor could soul intelligibly be said to get out of the way of 'perishing of soul', i.e. its own perishing.

O'Brien (op.cit. 102) says that death is here thought of in sense (i): 'This death is obviously death in the old sense, the death that forces the separation of the soul from the body'. But the old sense of 'death' seems inapplicable here. For in that sense the soul could just as well be said to 'die' as could the body. There would be no case for suggesting that the body 'dies' in that sense, but the soul does not. Nor, in the old sense of 'death', could the soul readily be said to 'get out of its way' (e7).

The difficulty here stems partly from the fact that the original definition of death (64c) had left it unclear what the proper subject of 'die' and 'be dead' is supposed to be (see on 64c2—9, p.86). Is it the whole man, the composite of body and soul, that is to be called 'dead', when its elements are separated from each other? Or may these elements themselves be said to 'die'? Death is almost invariably attributed either to the man (e.g. 70a3—4, b3, 80c2), or to a personal subject (as at 59a7, 87a4, 115d9), or to animate things (72c5—d3). 'Die' is occasionally used of the soul alone (77d3—4, 84b2, 88a6, 88d8). Only here is it used of the body, and Socrates avoids saying that the body 'dies' by referring to it as 'the mortal part'.

At 106e9—107al Socrates moves from the claim that 'soul' is immortal and imperishable to the claim that 'our souls' are in Hades. Once again the conclusion is drawn in terms of individual souls (see on 70c8-d6). Whether the shift from 'soul' to 'our souls' is justified will depend upon how the former is interpreted (see on 105c9—dl2). At some point in the argument it must be supposed that Socrates is talking of soul as a particular, not merely a Form. But if 'soul' means merely 'soul-stuff, the transition to 'our souls' might be questioned (see on 64e4—65a3, p.90).

107a2-bl0. What are 'the initial hypotheses' (b5), that Socrates tells his listeners they must explore further? It can hardly be doubted that they consist in, or at least include, the Theory of Forms. 'Hypotheses' (plural) could refer to the Theory alone, the positing of each Form being thought of as a separate hypothesis. In referring to them here, Socrates is recognizing that the whole case for immortality has rested upon postulates that are in need of further support. See on 65d4-e5 (p.97) and 101c9-102a9 (p.191).

The words 'you will, I believe, follow the argument to the furthest point to which man can follow it up' (b7—8) mean that the listeners will follow 'as far as is humanly possible'. They do not imply any reservations on Socrates' part as to the soundness of the argument, but simply echo Simmias' sentiments about 'human weakness' (bl). These are in keeping with his remarks at 85c—d about the limits of human reason. Cf. also 66e—67a.

4. MYTH (107cl-115a8)

The myth that now follows gives a speculative picture of the afterlife and judgement of departed souls. It includes a scientific theory as to the nature of the earth. And, by contrasting our earth with 'the true earth' above, it symbolizes the distinction between the sensible world and the world of Forms.

107cl-d2. Socrates here stresses the special need for 'care of the soul' in view of its immortality. K. Dorter (.Dialogue 1970, 574) takes the passage to mean that Socrates, by giving his listeners faith in immortality, deters them from 'nihilistic immorality'. But

Socrates does not say at 107c6—8 that if there were no afterlife, everything in this life would be permitted. He says that if death were the end, it would be a godsend for the wicked, since it would rid them of their wickedness. Far from suggesting that reward, or avoidance of punishment, in the next life is the reason for being good in this one, he implies that wickedness is burdensome in itself, whatever may happen after death. This is consistent with Plato's defence of justice in Republic ii—ix, which is independent of eternal rewards and punishments. See on 69a6—c3 (p.103), 81d6— 82d8.

At 107c3 the phrase 'what we call "life"' implies that ordinary usage wrongly restricts 'life' to the period of incarnation, whereas the soul 'lives', properly speaking, for ever. Plato often suggests that words are systematically misapplied and ordinary ways of speaking can mislead us. See 115e4—6. For other examples, cf.60b4, 64d3, 68c5—9, 71b7—8, 73b5, 75e5, 76a6-7,82bl, 86d3,95d4,96a7-8, 99b2—6, 112c2. See also Symposium 205b-d, Republic 493a-c.

108c5—109a8. The geophysical theories that follow include an account of the shape, position, surface, and inner structure of the earth, and a description of its seas, rivers, and volcanoes. Some of these subjects had been touched on by Socrates earlier (97d—98a, 99b—c). Here he shows more knowledge of them than his professed ineptitude for natural science might have suggested.

The identity of 'Glaucus' (108d4) is uncertain. But 'the skill of Glaucus' seems to have been proverbial for 'a great scientist' (Hackforth).

At 108e5 the translation 'round', rather than 'spherical', avoids prejudging the question whether the earth is, in fact, thought of as spherical. For the traditional interpretation see Bluck, 135, 200—1. For the alternatives see J. S. Morrison, Phronesis 1959, 101—19 (hemi-spherical), and T. G. Rosenmeyer, C.Q. 1956, 193-7, Phronesis 1959, 71—2 (disk-shaped), opposed by W. M. Calder, Phronesis 1958, 121-5.

113dl—114c8. Punishments in the afterlife are represented as purgatorial (d7—8), not vindictive (cf. E. R. Dodds, ed. Gorgias 525b 1—526d2). Only incurable malefactors are consigned to Tartarus for ever (el—6). Their punishment is a deterrent to other souls (cf. Republic 615c—616a), a belief which, as Dodds points out, makes sense only if the doctrine of rebirth is presupposed.

At 114a8—b5 souls are depicted as begging forgiveness for earthly misdeeds, which they must, presumably, remember having committed. Personal survival is, indeed, often held to entail the persistence of at least some memories from the life before death. But there has been nothing in the foregoing arguments for immort­ality to suggest that the discarnate soul remembers anything from its former life, or that it is capable of such states as penitence (al). See on 64e4-65a3 (p.90).

The idea that fully purified souls will live 'bodiless' (c3-4) for the rest of time has appeared before (81 a9), and implies that these souls will be altogether immune from rebirth. Strictly, this is incompatible with the conception of the soul's endless alternation between incarnate and discarnate states, which was posited at 77d4. For 'bodiless' existence cf.76cl2 and see on 76cl 1—13.

114dl—115a3. The status of the myth is indicated at 114d 1-6: exact knowledge of the afterlife is disclaimed, but belief in 'either this or something like it' is said to be 'a noble risk'. Such myths are, as E. R. Dodds has said (ed. Gorgias 523a2), 'a prolongation into the unknown of the lines established by philosophical argument'. They go far beyond anything the argument has suggested regarding the experience of the discarnate soul, and do not lend themselves to logical analysis.

'One should repeat such things to oneself like a spell' (d6—7). This recalls Socrates' earlier injunction to 'sing spells' to charm away the fear of death (77e8—9). But the myth just concluded would be more likely, one would think, to have the opposite effect on anyone who repeated it to himself, unless his conscience was unusually clear. Cf. Republic 330d4-331al.

114d8—115al should not be taken to imply that reward after death is the sole reason for practising the virtues mentioned here. See on 69a6-c3 (p.103), 81d6-82d8, 107cl-d2.

5. SOCRATES'DEATH (115bl —118al7)

Socrates gives final directions to his friends, and drinks the poison. His death follows.

115bl—116al. This passage contains some striking expressions of the idea that the soul is the 'true self. See on 64e4—65a3 (p.88).

117e3-118a4. For the action of the poison see Burnet, 149-50, and C. Gill, C.Q. 1973, 25-8. Gill argues that Plato's description of the symptoms is highly selective, and that the whole account is designed to represent the conception of soul that has been advocated in the discussion.

118a7—8. The offering of a cock to Asclepius is sometimes supposed to be for healing Socrates of the sickness of human life. He might refer to himself in the plural—cf.l 16d4. But the idea that life is a sickness, although once attributed to Cebes as part of an objection (95dl-4), is nowhere espoused by Socrates, and is hardly compatible with 90e2—91al. It is simpler to take the words as referring to an actual debt, incurred in some connection unknown. They are in keeping with the tribute to Socrates in the closing lines.

NOTES ON THE TEXT AND TRANSLATION

The translation 'prison' at 62b4 fits the theme of the soul's imprisonment in the body, which runs all through the dialogue (cf., e.g., 67dl—2, 82e—83a), and which is symbolized in its dramatic setting. If the meaning is 'guard' or 'garrison', the incarnate soul will be thought of as engaged in a military 'watch'. For a thorough discussion see Loriaux's note, and J. C. G. Strachan, C.Q. 1970, 216-20.

At 62cl, and occasionally elsewhere, ovkovv has been translated as if accented oHkovv. Cf., e.g., 71e4, 81a4, 83dl, and see J. D. Denniston, G.P. 432.

Or, taking fiaXkop with 6epnalveo6cu (63d7—8), 'people get overheated through talking'.

At 64a6, and generally, reOvavat has been translated 'be dead', despite Burnet's view, endorsed by Loriaux (50), that it may properly be translated 'die'. Burnet says (note on 62a5) that anodvyoiceiv lays stress on the process of dying, of which redvavai is the completion. However, anoOvrioKew will later be used (71c-e) to stress the process of dying, by contrast, rather, with the state of being dead. 'Die' is, no doubt, required in some of the passages cited by Burnet. In the Phaedo, however, it seems needed only at 62c3 and 81al. 'Be dead' is preferable at 62a5, 64c5 (in view of elvai at 64c8), 67e2 and 67e5, as well as in the present passage. It is clearly required for the opposition between to Tedvavai and to fr)i> at 71cl—5 and in the ensuing argument. Cf.77d3, and see on 71d5—e3.

The translation italicizes 'just' at 65d4 (and 'beautiful' and 'good' at 65d7) instead of translating avro, whose function is to mark a Form-referring use of the adjective. It is not_clear whether Sixaiov avro should be taken as complement of eivcu, as in the translation, or as subject: 'do we say that fust is something, or (that it is) nothing?' (cf.64c2, 102b 1). See also on 74a9-bl and 100b 1- c8.

Loriaux (85, 204) interprets 65dl 1—12 as if it meant 'Did you lay hold of them by some perception other than those that come by way of the body?' Read thus, the question would be properly answered 'Yes', and apprehension of the Forms would be viewed as a kind of intellectual 'perception'. However, if, as Loriaux also says (note on rqv o\ptv, p.87), aXXy tu>1 aioOfioei is echoed by aKkqv aioOr)ow at 65e8—66a 1, this interpretation is impossible. Moreover, apa... Oecopeirat (65el—2) seems to be a reformulation of the question being asked here, which must be parallel to that already asked at 65d9 for the eyes, and expecting the answer 'No'. For the meaning of aioBfioei on this view, see on 65bl—7.

More literally (65dl3—el): 'that which each one happens to be', taking o as complement of &>, and the whole clause as object of Aeyco. Note that the meaning of 'be', since the use is 'incomplete', cannot be existential here. This supports the interpretation of similar phrases adopted at 78d4—5. See note 31. obaia is used here with reference to the 'being' of individual Forms—cf. 101 c3. Elsewhere (76d9, 77a2, 78dl, 92d9) it is used more broadly for the whole domain of Forms. See C. H. Kahn, V.B.A.G. 460.

What does ctbrCov to akr\Qe0TaT0v at 65el—2 mean? Hackforth translates 'the full truth of them', and Bluck 'the truth about them'. It is not easy to get these meanings from the Greek. But to take avTtov as a partitive genitive would give an unsuitable sense, since (i) there can be no intention to limit the question being asked to 'the most real of the Forms', and (ii) there is no suggestion in the Phaedo that any one Form is more real than the others.

The translation retains t-qv at 65e7 and Tim at 65e8. Burnet's proposal to read rtc' at 65e7, and delete Tiva at 65e8, seems neither necessary nor even likely, if 6\piv and aSKrjp aiad-qaiv echo totc b^daXpok (65d9) and aXkri tivI aio9f}oei (65dll) respectively. The meaning of o\piv will depend upon the reading: t-qv o\j/w would be more naturally taken as 'the sense of sight', whereas tu>' dipiv could only mean 'any visual experience'. akkr)p alodrjow could mean, whether or not nud is retained, 'another sense experi­ence' or 'another sense': either of these might be said to be 'dragged in'.

The translation retains pern... onexjjei at 66b3—4, and takes eMpepew with J. E. Harry (C.R. 1909, 218-21) to mean 'lead astray', understanding the path in question to be the body. Philosophers will then be said to recognize that it side-tracks them in their quest for truth, and will thus be answering the question posed at 65a9—bl in the light of the intervening argument.